518 U.S. 515 (1996)


No. 94-1941.United States Supreme Court.

Argued January 17, 1996.Decided June 26, 1996.[*]CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FOURTH CIRCUIT517*517 517*517 518*518 Ginsburg, J.,delivered the opinion of the Court, in which Stevens, O’Connor, Kennedy, Souter, and Breyer, JJ., joined. Rehnquist, C. J., filed an opinion concurring in the judgment, post, p. 558. Scalia, J., filed a dissenting opinion, post, p. 566. Thomas, J., took no part in the consideration or decision of the case.

Paul Bender argued the cause for the United States in both cases. With him on the briefs were Solicitor General Days, Assistant Attorney General Patrick, Cornelia T. L. Pillard, Jessica Dunsay Silver, and Thomas E. Chandler.

Theodore B. Olson argued the cause and filed briefs for respondents in No. 94-1941 and petitioners in No. 94-2107. With him on the briefs were James S. Gilmore III, Attorney General of Virginia, William H. Hurd, Deputy Attorney General, Thomas G. Hungar, D. Jarrett Arp, Robert H. Patterson, Jr., Anne Marie Whittemore, William G. Broaddus, J. William Boland, Griffin B. Bell, and William A. Clineburg, Jr.[†]

519*519 Justice Ginsburg, delivered the opinion of the Court.

Virginia’s public institutions of higher learning include an incomparable military college, Virginia Military Institute (VMI). The United States maintains that the Constitution’s equal protection guarantee precludes Virginia from reserving exclusively to men the unique educational opportunities VMI affords. We agree.

520*520 I

Founded in 1839, VMI is today the sole single-sex school among Virginia’s 15 public institutions of higher learning. VMI’s distinctive mission is to produce “citizen-soldiers,” men prepared for leadership in civilian life and in military service. VMI pursues this mission through pervasive training of a kind not available anywhere else in Virginia. Assigning prime place to character development, VMI uses an “adversative method” modeled on English public schools and once characteristic of military instruction. VMI constantly endeavors to instill physical and mental discipline in its cadets and impart to them a strong moral code. The school’s graduates leave VMI with heightened comprehension of their capacity to deal with duress and stress, and a large sense of accomplishment for completing the hazardous course.

VMI has notably succeeded in its mission to produce leaders; among its alumni are military generals, Members of Congress, and business executives. The school’s alumni overwhelmingly perceive that their VMI training helped them to realize their personal goals. VMI’s endowment reflects the loyalty of its graduates; VMI has the largest per-student endowment of all public undergraduate institutions in the Nation.

Neither the goal of producing citizen-soldiers nor VMI’s implementing methodology is inherently unsuitable to women. And the school’s impressive record in producing leaders has made admission desirable to some women. Nevertheless, Virginia has elected to preserve exclusively for men the advantages and opportunities a VMI education affords.



From its establishment in 1839 as one of the Nation’s first state military colleges, see 1839 Va. Acts, ch. 20, VMI has remained financially supported by Virginia and “subject to 521*521 the control of the [Virginia] General Assembly,” Va. Code Ann. § 23-92 (1993). First southern college to teach engineering and industrial chemistry, see H. Wise, Drawing Out the Man: The VMI Story 13 (1978) (The VMI Story), VMI once provided teachers for the Commonwealth’s schools, see 1842 Va. Acts, ch. 24, § 2 (requiring every cadet to teach in one of the Commonwealth’s schools for a 2-year period).[1] Civil War strife threatened the school’s vitality, but a resourceful superintendent regained legislative support by highlighting “VMI’s great potential[,] through its technical know-how,” to advance Virginia’s postwar recovery. The VMI Story 47.

VMI today enrolls about 1,300 men as cadets.[2] Its academic offerings in the liberal arts, sciences, and engineering are also available at other public colleges and universities in Virginia. But VMI’s mission is special. It is the mission of the school

“`to produce educated and honorable men, prepared for the varied work of civil life, imbued with love of learning, confident in the functions and attitudes of leadership, possessing a high sense of public service, advocates of the American democracy and free enterprise system, and ready as citizen-soldiers to defend their country in 522*522 time of national peril.’ ” 766 F. Supp. 1407, 1425 (WD Va. 1991) (quoting Mission Study Committee of the VMI Board of Visitors, Report, May 16, 1986).

In contrast to the federal service academies, institutions maintained “to prepare cadets for career service in the armed forces,” VMI’s program “is directed at preparation for both military and civilian life”; “[o]nly about 15% of VMI cadets enter career military service.” 766 F. Supp., at 1432.

VMI produces its “citizen-soldiers” through “an adversative, or doubting, model of education” which features “[p]hysical rigor, mental stress, absolute equality of treatment, absence of privacy, minute regulation of behavior, and indoctrination in desirable values.” Id., at 1421. As one Commandant of Cadets described it, the adversative method “`dissects the young student,’ ” and makes him aware of his “`limits and capabilities,’ ” so that he knows “`how far he can go with his anger, . . . how much he can take under stress, . . . exactly what he can do when he is physically exhausted.’ ” Id. , at 1421-1422 (quoting Col. N. Bissell).

VMI cadets live in spartan barracks where surveillance is constant and privacy nonexistent; they wear uniforms, eat together in the mess hall, and regularly participate in drills. Id. , at 1424, 1432. Entering students are incessantly exposed to the rat line, “an extreme form of the adversative model,” comparable in intensity to Marine Corps boot camp. Id. , at 1422. Tormenting and punishing, the rat line bonds new cadets to their fellow sufferers and, when they have completed the 7-month experience, to their former tormentors. Ibid.

VMI’s “adversative model” is further characterized by a hierarchical “class system” of privileges and responsibilities, a “dyke system” for assigning a senior class mentor to each entering class “rat,” and a stringently enforced “honor code,” which prescribes that a cadet “`does not lie, cheat, steal nor tolerate those who do.’ ” Id. , at 1422-1423.

523*523 VMI attracts some applicants because of its reputation as an extraordinarily challenging military school, and “because its alumni are exceptionally close to the school.” Id. , at 1421. “[W]omen have no opportunity anywhere to gain the benefits of [the system of education at VMI].” Ibid.


In 1990, prompted by a complaint filed with the Attorney General by a female high-school student seeking admission to VMI, the United States sued the Commonwealth of Virginia and VMI, alleging that VMI’s exclusively male admission policy violated the Equal Protection Clause of the Fourteenth Amendment. Id. , at 1408.[3] Trial of the action consumed six days and involved an array of expert witnesses on each side. Ibid.

In the two years preceding the lawsuit, the District Court noted, VMI had received inquiries from 347 women, but had responded to none of them. Id. , at 1436. “[S]ome women, at least,” the court said, “would want to attend the school if they had the opportunity.” Id. , at 1414. The court further recognized that, with recruitment, VMI could “achieve at least 10% female enrollment”—”a sufficient `critical mass’ to provide the female cadets with a positive educational experience.” Id. , at 1437-1438. And it was also established that “some women are capable of all of the individual activities required of VMI cadets.” Id. , at 1412. In addition, experts agreed that if VMI admitted women, “the VMI ROTC experience would become a better training program from the perspective of the armed forces, because it would provide training in dealing with a mixed-gender army.” Id. , at 1441.

The District Court ruled in favor of VMI, however, and rejected the equal protection challenge pressed by the United States. That court correctly recognized that Mississippi Univ. for Women v. Hogan, 458 U. S. 718 (1982), was 524*524 the closest guide. 766 F. Supp., at 1410. There, this Court underscored that a party seeking to uphold government action based on sex must establish an “exceedingly persuasive justification” for the classification. Mississippi Univ. for Women, 458 U. S., at 724 (internal quotation marks omitted). To succeed, the defender of the challenged action must show “at least that the classification serves important governmental objectives and that the discriminatory means employed are substantially related to the achievement of those objectives.” Ibid. (internal quotation marks omitted).

The District Court reasoned that education in “a singlegender environment, be it male or female,” yields substantial benefits. 766 F. Supp., at 1415. VMI’s school for men brought diversity to an otherwise coeducational Virginia system, and that diversity was “enhanced by VMI’s unique method of instruction.” Ibid. If single-gender education for males ranks as an important governmental objective, it becomes obvious, the District Court concluded, that the only means of achieving the objective “is to exclude women from the all-male institution—VMI.” Ibid.

“Women are [indeed] denied a unique educational opportunity that is available only at VMI,” the District Court acknowledged. Id. , at 1432. But “[VMI’s] single-sex status would be lost, and some aspects of the [school’s] distinctive method would be altered,” if women were admitted, id. , at 1413: “Allowance for personal privacy would have to be made,” id. , at 1412; “[p]hysical education requirements would have to be altered, at least for the women,” id. , at 1413; the adversative environment could not survive unmodified, id. , at 1412-1413. Thus, “sufficient constitutional justification” had been shown, the District Court held, “for continuing [VMI’s] single-sex policy.” Id. , at 1413.

The Court of Appeals for the Fourth Circuit disagreed and vacated the District Court’s judgment. The appellate court held: “The Commonwealth of Virginia has not . . . advanced any state policy by which it can justify its determination, 525*525 under an announced policy of diversity, to afford VMI’s unique type of program to men and not to women.” 976 F. 2d 890, 892 (1992).

The appeals court greeted with skepticism Virginia’s assertion that it offers single-sex education at VMI as a facet of the Commonwealth’s overarching and undisputed policy to advance “autonomy and diversity.” The court underscored Virginia’s nondiscrimination commitment: “`[I]t is extremely important that [colleges and universities] deal with faculty, staff, and students without regard to sex, race, or ethnic origin. ` ” Id. , at 899 (quoting 1990 Report of the Virginia Commission on the University of the 21st Century). “That statement,” the Court of Appeals said, “is the only explicit one that we have found in the record in which the Commonwealth has expressed itself with respect to gender distinctions.” 976 F. 2d, at 899. Furthermore, the appeals court observed, in urging “diversity” to justify an all-male VMI, the Commonwealth had supplied “no explanation for the movement away from [single-sex education] in Virginia by public colleges and universities.” Ibid. In short, the court concluded, “[a] policy of diversity which aims to provide an array of educational opportunities, including single-gender institutions, must do more than favor one gender.” Ibid.

The parties agreed that “some women can meet the physical standards now imposed on men,” id. , at 896, and the court was satisfied that “neither the goal of producing citizen soldiers nor VMI’s implementing methodology is inherently unsuitable to women,” id. , at 899. The Court of Appeals, however, accepted the District Court’s finding that “at least these three aspects of VMI’s program—physical training, the absence of privacy, and the adversative approach—would be materially affected by coeducation.” Id. , at 896-897. Remanding the case, the appeals court assigned to Virginia, in the first instance, responsibility for selecting a remedial course. The court suggested these options for the Commonwealth: Admit women to VMI; establish parallel institutions 526*526 or programs; or abandon state support, leaving VMI free to pursue its policies as a private institution. Id. , at 900. In May 1993, this Court denied certiorari. See 508 U. S. 946; see also ibid. (opinion of Scalia, J., noting the interlocutory posture of the litigation).


In response to the Fourth Circuit’s ruling, Virginia proposed a parallel program for women: Virginia Women’s Institute for Leadership (VWIL). The 4-year, state-sponsored undergraduate program would be located at Mary Baldwin College, a private liberal arts school for women, and would be open, initially, to about 25 to 30 students. Although VWIL would share VMI’s mission—to produce “citizensoldiers”—the VWIL program would differ, as does Mary Baldwin College, from VMI in academic offerings, methods of education, and financial resources. See 852 F. Supp. 471, 476-477 (WD Va. 1994).

The average combined SAT score of entrants at Mary Baldwin is about 100 points lower than the score for VMI freshmen. See id., at 501. Mary Baldwin’s faculty holds “significantly fewer Ph. D.’s than the faculty at VMI,” id. , at 502, and receives significantly lower salaries, see Tr. 158 (testimony of James Lott, Dean of Mary Baldwin College), reprinted in 2 App. in Nos. 94-1667 and 94-1717 (CA4) (hereinafter Tr.). While VMI offers degrees in liberal arts, the sciences, and engineering, Mary Baldwin, at the time of trial, offered only bachelor of arts degrees. See 852 F. Supp., at 503. A VWIL student seeking to earn an engineering degree could gain one, without public support, by attending Washington University in St. Louis, Missouri, for two years, paying the required private tuition. See ibid.

Experts in educating women at the college level composed the Task Force charged with designing the VWIL program; Task Force members were drawn from Mary Baldwin’s own faculty and staff. Id. , at 476. Training its attention on methods of instruction appropriate for “most women,” the 527*527 Task Force determined that a military model would be “wholly inappropriate” for VWIL. Ibid.; see 44 F. 3d 1229, 1233 (CA4 1995).

VWIL students would participate in ROTC programs and a newly established, “largely ceremonial” Virginia Corps of Cadets, id. , at 1234, but the VWIL House would not have a military format, 852 F. Supp., at 477, and VWIL would not require its students to eat meals together or to wear uniforms during the schoolday, id., at 495. In lieu of VMI’s adversative method, the VWIL Task Force favored “a cooperative method which reinforces self-esteem.” Id., at 476. In addition to the standard bachelor of arts program offered at Mary Baldwin, VWIL students would take courses in leadership, complete an off-campus leadership externship, participate in community service projects, and assist in arranging a speaker series. See 44 F. 3d, at 1234.

Virginia represented that it will provide equal financial support for in-state VWIL students and VMI cadets, 852 F. Supp., at 483, and the VMI Foundation agreed to supply a $5.4625 million endowment for the VWIL program, id., at 499. Mary Baldwin’s own endowment is about $19 million; VMI’s is $131 million. Id., at 503. Mary Baldwin will add $35 million to its endowment based on future commitments; VMI will add $220 million. Ibid. The VMI Alumni Association has developed a network of employers interested in hiring VMI graduates. The Association has agreed to open its network to VWIL graduates, id., at 499, but those graduates will not have the advantage afforded by a VMI degree.


Virginia returned to the District Court seeking approval of its proposed remedial plan, and the court decided the plan met the requirements of the Equal Protection Clause. Id. , at 473. The District Court again acknowledged evidentiary support for these determinations: “[T]he VMI methodology could be used to educate women and, in fact, some 528*528 women . . . may prefer the VMI methodology to the VWIL methodology.” Id. , at 481. But the “controlling legal principles,” the District Court decided, “do not require the Commonwealth to provide a mirror image VMI for women.” Ibid. The court anticipated that the two schools would “achieve substantially similar outcomes.” Ibid. It concluded: “If VMI marches to the beat of a drum, then Mary Baldwin marches to the melody of a fife and when the march is over, both will have arrived at the same destination.” Id. , at 484.

A divided Court of Appeals affirmed the District Court’s judgment. 44 F. 3d 1229 (CA4 1995). This time, the appellate court determined to give “greater scrutiny to the selection of means than to the [Commonwealth’s] proffered objective.” Id. , at 1236. The official objective or purpose, the court said, should be reviewed deferentially. Ibid. Respect for the “legislative will,” the court reasoned, meant that the judiciary should take a “cautious approach,” inquiring into the “legitima[cy]” of the governmental objective and refusing approval for any purpose revealed to be “pernicious.” Ibid.

“[P]roviding the option of a single-gender college education may be considered a legitimate and important aspect of a public system of higher education,” the appeals court observed, id., at 1238; that objective, the court added, is “not pernicious,” id., at 1239. Moreover, the court continued, the adversative method vital to a VMI education “has never been tolerated in a sexually heterogeneous environment.” Ibid. The method itself “was not designed to exclude women,” the court noted, but women could not be accommodated in the VMI program, the court believed, for female participation in VMI’s adversative training “would destroy. . . any sense of decency that still permeates the relationship between the sexes.” Ibid.

Having determined, deferentially, the legitimacy of Virginia’s purpose, the court considered the question of means. 529*529 Exclusion of “men at Mary Baldwin College and women at VMI,” the court said, was essential to Virginia’s purpose, for without such exclusion, the Commonwealth could not “accomplish [its] objective of providing single-gender education.” Ibid.

The court recognized that, as it analyzed the case, means merged into end, and the merger risked “bypass[ing] any equal protection scrutiny.” Id., at 1237. The court therefore added another inquiry, a decisive test it called “substantive comparability.” Ibid. The key question, the court said, was whether men at VMI and women at VWIL would obtain “substantively comparable benefits at their institution or through other means offered by the [S]tate.” Ibid. Although the appeals court recognized that the VWIL degree “lacks the historical benefit and prestige” of a VMI degree, it nevertheless found the educational opportunities at the two schools “sufficiently comparable.” Id., at 1241.

Senior Circuit Judge Phillips dissented. The court, in his judgment, had not held Virginia to the burden of showing an “`exceedingly persuasive [justification]’ ” for the Commonwealth’s action. Id., at 1247 (quoting Mississippi Univ. for Women, 458 U. S., at 724). In Judge Phillips’ view, the court had accepted “rationalizations compelled by the exigencies of this litigation,” and had not confronted the Commonwealth’s “actual overriding purpose.” 44 F. 3d, at 1247. That purpose, Judge Phillips said, was clear from the historical record; it was “not to create a new type of educational opportunity for women, . . . nor to further diversify the Commonwealth’s higher education system[,] . . . but [was] simply . . . to allow VMI to continue to exclude women in order to preserve its historic character and mission.” Ibid.

Judge Phillips suggested that the Commonwealth would satisfy the Constitution’s equal protection requirement if it “simultaneously opened single-gender undergraduate institutions having substantially comparable curricular and extra-curricular programs, funding, physical plant, administration 530*530 and support services, and faculty and library resources.” Id., at 1250. But he thought it evident that the proposed VWIL program, in comparison to VMI, fell “far short . . . from providing substantially equal tangible and intangible educational benefits to men and women.” Ibid.

The Fourth Circuit denied rehearing en banc. 52 F. 3d 90 (1995). Circuit Judge Motz, joined by Circuit Judges Hall, Murnaghan, and Michael, filed a dissenting opinion.[4] Judge Motz agreed with Judge Phillips that Virginia had not shown an “`exceedingly persuasive justification’ ” for the disparate opportunities the Commonwealth supported. Id., at 92 (quoting Mississippi Univ. for Women, 458 U. S., at 724). She asked: “[H]ow can a degree from a yet to be implemented supplemental program at Mary Baldwin be held `substantively comparable’ to a degree from a venerable Virginia military institution that was established more than 150 years ago?” 52 F. 3d, at 93. “Women need not be guaranteed equal `results,’ ” Judge Motz said, “but the Equal Protection Clause does require equal opportunity . . . [and] that opportunity is being denied here.” Ibid.


The cross-petitions in this suit present two ultimate issues. First, does Virginia’s exclusion of women from the educational opportunities provided by VMI—extraordinary opportunities for military training and civilian leadership development—deny to women “capable of all of the individual activities required of VMI cadets,” 766 F. Supp., at 1412, the equal protection of the laws guaranteed by the Fourteenth Amendment? Second, if VMI’s “unique” situation, id. , at 1413—as Virginia’s sole single-sex public institution of 531*531 higher education—offends the Constitution’s equal protection principle, what is the remedial requirement?


We note, once again, the core instruction of this Court’s pathmarking decisions in J. E. B. v. Alabama ex rel. T. B., 511 U. S. 127, 136-137, and n. 6 (1994), and Mississippi Univ. for Women, 458 U. S., at 724 (internal quotation marks omitted): Parties who seek to defend gender-based government action must demonstrate an “exceedingly persuasive justification” for that action.

Today’s skeptical scrutiny of official action denying rights or opportunities based on sex responds to volumes of history. As a plurality of this Court acknowledged a generation ago, “our Nation has had a long and unfortunate history of sex discrimination.” Frontiero v. Richardson, 411 U. S. 677, 684 (1973). Through a century plus three decades and more of that history, women did not count among voters composing “We the People”;[5] not until 1920 did women gain a constitutional right to the franchise. Id., at 685. And for a half century thereafter, it remained the prevailing doctrine that government, both federal and state, could withhold from women opportunities accorded men so long as any “basis in reason” could be conceived for the discrimination. See, e. g., Goesaert v. Cleary, 335 U. S. 464, 467 (1948) (rejecting challenge of female tavern owner and her daughter to Michigan law denying bartender licenses to females—except for wives and daughters of male tavern owners; Court would not “give ear” to the contention that “an unchivalrous desire of male 532*532 bartenders to . . . monopolize the calling” prompted the legislation).

In 1971, for the first time in our Nation’s history, this Court ruled in favor of a woman who complained that her State had denied her the equal protection of its laws. Reed v. Reed, 404 U. S. 71, 73 (holding unconstitutional Idaho Code prescription that, among “`several persons claiming and equally entitled to administer [a decedent’s estate], males must be preferred to females’ “). Since Reed, the Court has repeatedly recognized that neither federal nor state government acts compatibly with the equal protection principle when a law or official policy denies to women, simply because they are women, full citizenship stature—equal opportunity to aspire, achieve, participate in and contribute to society based on their individual talents and capacities. See, e. g., Kirchberg v. Feenstra, 450 U. S. 455, 462-463 (1981) (affirming invalidity of Louisiana law that made husband “head and master” of property jointly owned with his wife, giving him unilateral right to dispose of such property without his wife’s consent); Stanton v. Stanton, 421 U. S. 7 (1975) (invalidating Utah requirement that parents support boys until age 21, girls only until age 18).

Without equating gender classifications, for all purposes, to classifications based on race or national origin,[6] the Court, in post-Reed decisions, has carefully inspected official action that closes a door or denies opportunity to women (or to men). See J. E. B. , 511 U. S., at 152 (Kennedy, J., concurring in judgment) (case law evolving since 1971 “reveal[s] a strong presumption that gender classifications are invalid”). To summarize the Court’s current directions for cases of official classification based on gender: Focusing on the differential 533*533 treatment or denial of opportunity for which relief is sought, the reviewing court must determine whether the proffered justification is “exceedingly persuasive.” The burden of justification is demanding and it rests entirely on the State. See Mississippi Univ. for Women, 458 U. S., at 724. The State must show “at least that the [challenged] classification serves `important governmental objectives and that the discriminatory means employed’ are `substantially related to the achievement of those objectives.’ ” Ibid. (quoting Wengler v. Druggists Mut. Ins. Co. , 446 U. S. 142, 150 (1980)). The justification must be genuine, not hypothesized or invented post hoc in response to litigation. And it must not rely on overbroad generalizations about the different talents, capacities, or preferences of males and females. See Weinberger v. Wiesenfeld, 420 U. S. 636, 643, 648 (1975); Califano v. Goldfarb, 430 U. S. 199, 223-224 (1977) (Stevens, J., concurring in judgment).

The heightened review standard our precedent establishes does not make sex a proscribed classification. Supposed “inherent differences” are no longer accepted as a ground for race or national origin classifications. See Loving v. Virginia, 388 U. S. 1 (1967). Physical differences between men and women, however, are enduring: “[T]he two sexes are not fungible; a community made up exclusively of one [sex] is different from a community composed of both.” Ballard v. United States, 329 U. S. 187, 193 (1946).

“Inherent differences” between men and women, we have come to appreciate, remain cause for celebration, but not for denigration of the members of either sex or for artificial constraints on an individual’s opportunity. Sex classifications may be used to compensate women “for particular economic disabilities [they have] suffered,” Califano v. Webster, 430 U. S. 313, 320 (1977) (per curiam), to “promot[e] equal employment opportunity,” see California Fed. Sav. & Loan Assn. v. Guerra, 479 U. S. 272, 289 (1987), to advance full development of the talent and capacities of our Nation’s people.[7] 534*534 But such classifications may not be used, as they once were, see Goesaert, 335 U. S., at 467, to create or perpetuate the legal, social, and economic inferiority of women.

Measuring the record in this case against the review standard just described, we conclude that Virginia has shown no “exceedingly persuasive justification” for excluding all women from the citizen-soldier training afforded by VMI. We therefore affirm the Fourth Circuit’s initial judgment, which held that Virginia had violated the Fourteenth Amendment’s Equal Protection Clause. Because the remedy proffered by Virginia—the Mary Baldwin VWIL program—does not cure the constitutional violation, i. e. , it does not provide equal opportunity, we reverse the Fourth Circuit’s final judgment in this case.


The Fourth Circuit initially held that Virginia had advanced no state policy by which it could justify, under equal protection principles, its determination “to afford VMI’s unique type of program to men and not to women.” 976 F. 2d, at 892. Virginia challenges that “liability” ruling and asserts two justifications in defense of VMI’s exclusion of 535*535 women. First, the Commonwealth contends, “single-sex education provides important educational benefits,” Brief for Cross-Petitioners 20, and the option of single-sex education contributes to “diversity in educational approaches,” id. , at 25. Second, the Commonwealth argues, “the unique VMI method of character development and leadership training,” the school’s adversative approach, would have to be modified were VMI to admit women. Id. , at 33-36 (internal quotation marks omitted). We consider these two justifications in turn.


Single-sex education affords pedagogical benefits to at least some students, Virginia emphasizes, and that reality is uncontested in this litigation.[8] Similarly, it is not disputed that diversity among public educational institutions can serve the public good. But Virginia has not shown that VMI was established, or has been maintained, with a view to diversifying, by its categorical exclusion of women, educational opportunities within the Commonwealth. In cases of this genre, our precedent instructs that “benign” justifications proffered in defense of categorical exclusions will not be accepted automatically; a tenable justification must describe actual state purposes, not rationalizations for actions 536*536 in fact differently grounded. See Wiesenfeld, 420 U. S., at 648, and n. 16 (“mere recitation of a benign [or] compensatory purpose” does not block “inquiry into the actual purposes” of government-maintained gender-based classifications); Goldfarb, 430 U. S., at 212-213 (rejecting government-proffered purposes after “inquiry into the actual purposes” (internal quotation marks omitted)).

Mississippi Univ. for Women is immediately in point. There the State asserted, in justification of its exclusion of men from a nursing school, that it was engaging in “educational affirmative action” by “compensat[ing] for discrimination against women.” 458 U. S., at 727. Undertaking a “searching analysis,” id. , at 728, the Court found no close resemblance between “the alleged objective” and “the actual purpose underlying the discriminatory classification,” id. , at 730. Pursuing a similar inquiry here, we reach the same conclusion.

Neither recent nor distant history bears out Virginia’s alleged pursuit of diversity through single-sex educational options. In 1839, when the Commonwealth established VMI, a range of educational opportunities for men and women was scarcely contemplated. Higher education at the time was considered dangerous for women;[9] reflecting 537*537 widely held views about women’s proper place, the Nation’s first universities and colleges—for example, Harvard in Massachusetts, William and Mary in Virginia—admitted only men. See E. Farello, A History of the Education of Women in the United States 163 (1970). VMI was not at all novel in this respect: In admitting no women, VMI followed the lead of the Commonwealth’s flagship school, the University of Virginia, founded in 1819.

“[N]o struggle for the admission of women to a state university,” a historian has recounted, “was longer drawn out, or developed more bitterness, than that at the University of Virginia.” 2 T. Woody, A History of Women’s Education in the United States 254 (1929) (History of Women’s Education). In 1879, the State Senate resolved to look into the possibility of higher education for women, recognizing that Virginia “`has never, at any period of her history,’ ” provided for the higher education of her daughters, though she “`has liberally provided for the higher education of her sons.’ ” Ibid. (quoting 10 Educ. J. Va. 212 (1879)). Despite this recognition, no new opportunities were instantly open to women.[10]

Virginia eventually provided for several women’s seminaries and colleges. Farmville Female Seminary became a public institution in 1884. See supra, at 521, n. 2. Two women’s schools, Mary Washington College and James Madison University, were founded in 1908; another, Radford University, was founded in 1910. 766 F. Supp., at 1418-1419. By the mid-1970’s, all four schools had become coeducational. Ibid.

Debate concerning women’s admission as undergraduates at the main university continued well past the century’s midpoint. Familiar arguments were rehearsed. If women 538*538 were admitted, it was feared, they “would encroach on the rights of men; there would be new problems of government, perhaps scandals; the old honor system would have to be changed; standards would be lowered to those of other coeducational schools; and the glorious reputation of the university, as a school for men, would be trailed in the dust.” 2 History of Women’s Education 255.

Ultimately, in 1970, “the most prestigious institution of higher education in Virginia,” the University of Virginia, introduced coeducation and, in 1972, began to admit women on an equal basis with men. See Kirstein v. Rector and Visitors of Univ. of Virginia, 309 F. Supp. 184, 186 (ED Va. 1970). A three-judge Federal District Court confirmed: “Virginia may not now deny to women, on the basis of sex, educational opportunities at the Charlottesville campus that are not afforded in other institutions operated by the [S]tate.” Id. , at 187.

Virginia describes the current absence of public single-sex higher education for women as “an historical anomaly.” Brief for Cross-Petitioners 30. But the historical record indicates action more deliberate than anomalous: First, protection of women against higher education; next, schools for women far from equal in resources and stature to schools for men; finally, conversion of the separate schools to coeducation. The state legislature, prior to the advent of this controversy, had repealed “[a]ll Virginia statutes requiring individual institutions to admit only men or women.” 766 F. Supp., at 1419. And in 1990, an official commission, “legislatively established to chart the future goals of higher education in Virginia,” reaffirmed the policy “`of affording broad access” while maintaining “autonomy and diversity.’ ” 976 F. 2d, at 898-899 (quoting Report of the Virginia Commission on the University of the 21st Century). Significantly, the commission reported:

“`Because colleges and universities provide opportunities for students to develop values and learn from role 539*539 models, it is extremely important that they deal with faculty, staff, and students without regard to sex, race, or ethnic origin.’ ” Id. , at 899 (emphasis supplied by Court of Appeals deleted).

This statement, the Court of Appeals observed, “is the only explicit one that we have found in the record in which the Commonwealth has expressed itself with respect to gender distinctions.” Ibid.

Our 1982 decision in Mississippi Univ. for Women prompted VMI to reexamine its male-only admission policy. See 766 F. Supp., at 1427-1428. Virginia relies on that reexamination as a legitimate basis for maintaining VMI’s single-sex character. See Reply Brief for Cross-Petitioners 6. A Mission Study Committee, appointed by the VMI Board of Visitors, studied the problem from October 1983 until May 1986, and in that month counseled against “change of VMI status as a single-sex college.” See 766 F. Supp., at 1429 (internal quotation marks omitted). Whatever internal purpose the Mission Study Committee served—and however well meaning the framers of the report—we can hardly extract from that effort any commonwealth policy evenhandedly to advance diverse educational options. As the District Court observed, the Committee’s analysis “primarily focuse[d] on anticipated difficulties in attracting females to VMI,” and the report, overall, supplied “very little indication of how th[e] conclusion was reached.” Ibid.

In sum, we find no persuasive evidence in this record that VMI’s male-only admission policy “is in furtherance of a state policy of `diversity.’ ” See 976 F. 2d, at 899. No such policy, the Fourth Circuit observed, can be discerned from the movement of all other public colleges and universities in Virginia away from single-sex education. See ibid. That court also questioned “how one institution with autonomy, but with no authority over any other state institution, can give effect to a state policy of diversity among institutions.” Ibid. A purpose genuinely to advance an array of educational 540*540 options, as the Court of Appeals recognized, is not served by VMI’s historic and constant plan—a plan to “affor[d] a unique educational benefit only to males.” Ibid. However “liberally” this plan serves the Commonwealth’s sons, it makes no provision whatever for her daughters. That is not equal protection.


Virginia next argues that VMI’s adversative method of training provides educational benefits that cannot be made available, unmodified, to women. Alterations to accommodate women would necessarily be “radical,” so “drastic,” Virginia asserts, as to transform, indeed “destroy,” VMI’s program. See Brief for Cross-Petitioners 34-36. Neither sex would be favored by the transformation, Virginia maintains: Men would be deprived of the unique opportunity currently available to them; women would not gain that opportunity because their participation would “eliminat[e] the very aspects of [the] program that distinguish [VMI] from . . . other institutions of higher education in Virginia.” Id., at 34.

The District Court forecast from expert witness testimony, and the Court of Appeals accepted, that coeducation would materially affect “at least these three aspects of VMI’s program—physical training, the absence of privacy, and the adversative approach.” 976 F. 2d, at 896-897. And it is uncontested that women’s admission would require accommodations, primarily in arranging housing assignments and physical training programs for female cadets. See Brief for Cross-Respondent 11, 29-30. It is also undisputed, however, that “the VMI methodology could be used to educate women.” 852 F. Supp., at 481. The District Court even allowed that some women may prefer it to the methodology a women’s college might pursue. See ibid. “[S]ome women, at least, would want to attend [VMI] if they had the opportunity,” the District Court recognized, 766 F. Supp., at 1414, and “some women,” the expert testimony established, “are 541*541 capable of all of the individual activities required of VMI cadets,” id., at 1412. The parties, furthermore, agree that “some women can meet the physical standards [VMI] now impose[s] on men.” 976 F. 2d, at 896. In sum, as the Court of Appeals stated, “neither the goal of producing citizen soldiers,” VMI’s raison d’être, “nor VMI’s implementing methodology is inherently unsuitable to women.” Id., at 899.

In support of its initial judgment for Virginia, a judgment rejecting all equal protection objections presented by the United States, the District Court made “findings” on “gender-based developmental differences.” 766 F. Supp., at 1434-1435. These “findings” restate the opinions of Virginia’s expert witnesses, opinions about typically male or typically female “tendencies.” Id., at 1434. For example, “[m]ales tend to need an atmosphere of adversativeness,” while “[f]emales tend to thrive in a cooperative atmosphere.” Ibid. “I’m not saying that some women don’t do well under [the] adversative model,” VMI’s expert on educational institutions testified, “undoubtedly there are some [women] who do”; but educational experiences must be designed “around the rule,” this expert maintained, and not “around the exception.” Ibid. (internal quotation marks omitted).

The United States does not challenge any expert witness estimation on average capacities or preferences of men and women. Instead, the United States emphasizes that time and again since this Court’s turning point decision in Reed v. Reed, 404 U. S. 71 (1971), we have cautioned reviewing courts to take a “hard look” at generalizations or “tendencies” of the kind pressed by Virginia, and relied upon by the District Court. See O’Connor, Portia’s Progress, 66 N. Y. U. L. Rev. 1546, 1551 (1991). State actors controlling gates to opportunity, we have instructed, may not exclude qualified individuals based on “fixed notions concerning the roles and abilities of males and females.” Mississippi Univ. for Women, 458 U. S., at 725; see J. E. B., 511 U. S., at 139, n. 11 (equal protection principles, as applied to gender classifications, mean 542*542 state actors may not rely on “overbroad” generalizations to make “judgments about people that are likely to . . . perpetuate historical patterns of discrimination”).

It may be assumed, for purposes of this decision, that most women would not choose VMI’s adversative method. As Fourth Circuit Judge Motz observed, however, in her dissent from the Court of Appeals’ denial of rehearing en banc, it is also probable that “many men would not want to be educated in such an environment.” 52 F. 3d, at 93. (On that point, even our dissenting colleague might agree.) Education, to be sure, is not a “one size fits all” business. The issue, however, is not whether “women—or men—should be forced to attend VMI”; rather, the question is whether the Commonwealth can constitutionally deny to women who have the will and capacity, the training and attendant opportunities that VMI uniquely affords. Ibid.

The notion that admission of women would downgrade VMI’s stature, destroy the adversative system and, with it, even the school,[11] is a judgment hardly proved,[12] a prediction 543*543 hardly different from other “self-fulfilling prophec[ies],” see Mississippi Univ. for Women, 458 U. S., at 730, once routinely used to deny rights or opportunities. When women first sought admission to the bar and access to legal education, concerns of the same order were expressed. For example, in 1876, the Court of Common Pleas of Hennepin County, Minnesota, explained why women were thought ineligible for the practice of law. Women train and educate the young, the court said, which

“forbids that they shall bestow that time (early and late) and labor, so essential in attaining to the eminence to which the true lawyer should ever aspire. It cannot therefore be said that the opposition of courts to the admission of females to practice . . . is to any extent the outgrowth of . . . `old fogyism[.]’ . . . [I]t arises rather from a comprehension of the magnitude of the responsibilities connected with the successful practice of law, and a desire to grade up the profession.” In re Application of Martha Angle Dorsett to Be Admitted to Practice as Attorney and Counselor at Law (Minn. C. P. Hennepin Cty., 1876), in The Syllabi, Oct. 21, 1876, pp. 5, 6 (emphasis added).

A like fear, according to a 1925 report, accounted for Columbia Law School’s resistance to women’s admission, although

“[t]he faculty . . . never maintained that women could not master legal learning . . . . No, its argument has been . . . more practical. If women were admitted to 544*544 the Columbia Law School, [the faculty] said, then the choicer, more manly and red-blooded graduates of our great universities would go to the Harvard Law School!” The Nation, Feb. 18, 1925, p. 173.

Medical faculties similarly resisted men and women as partners in the study of medicine. See R. Morantz-Sanchez, Sympathy and Science: Women Physicians in American Medicine 51-54, 250 (1985); see also M. Walsh, “Doctors Wanted: No Women Need Apply” 121-122 (1977) (quoting E. Clarke, Medical Education of Women, 4 Boston Med. & Surg. J. 345, 346 (1869) (“`God forbid that I should ever see men and women aiding each other to display with the scalpel the secrets of the reproductive system . . . .’ “)); cf. supra, at 536-537, n. 9. More recently, women seeking careers in policing encountered resistance based on fears that their presence would “undermine male solidarity,” see F. Heidensohn, Women in Control? 201 (1992); deprive male partners of adequate assistance, see id. , at 184-185; and lead to sexual misconduct, see C. Milton et al., Women in Policing 32-33 (1974). Field studies did not confirm these fears. See Heidensohn, supra, at 92-93; P. Bloch & D. Anderson, Policewomen on Patrol: Final Report (1974).

Women’s successful entry into the federal military academies,[13] and their participation in the Nation’s military forces,[14] indicate that Virginia’s fears for the future of VMI 545*545 may not be solidly grounded.[15] The Commonwealth’s justification for excluding all women from “citizen-soldier” training for which some are qualified, in any event, cannot rank as “exceedingly persuasive,” as we have explained and applied that standard.

Virginia and VMI trained their argument on “means” rather than “end,” and thus misperceived our precedent. Single-sex education at VMI serves an “important governmental objective,” they maintained, and exclusion of women is not only “substantially related,” it is essential to that objective. By this notably circular argument, the “straightforward” test Mississippi Univ. for Women described, see 458 U. S., at 724-725, was bent and bowed.

The Commonwealth’s misunderstanding and, in turn, the District Court’s, is apparent from VMI’s mission: to produce “citizen-soldiers,” individuals

“`imbued with love of learning, confident in the functions and attitudes of leadership, possessing a high sense of public service, advocates of the American democracy and free enterprise system, and ready . . . to defend their country in time of national peril.’ ” 766 F. Supp., at 1425 (quoting Mission Study Committee of the VMI Board of Visitors, Report, May 16, 1986).

Surely that goal is great enough to accommodate women, who today count as citizens in our American democracy equal in stature to men. Just as surely, the Commonwealth’s 546*546 great goal is not substantially advanced by women’s categorical exclusion, in total disregard of their individual merit, from the Commonwealth’s premier “citizen-soldier” corps.[16] Virginia, in sum, “has fallen far short of establishing the `exceedingly persuasive justification,’ ” Mississippi Univ. for Women, 458 U. S., at 731, that must be the solid base for any gender-defined classification.


In the second phase of the litigation, Virginia presented its remedial plan—maintain VMI as a male-only college and create VWIL as a separate program for women. The plan met District Court approval. The Fourth Circuit, in turn, deferentially reviewed the Commonwealth’s proposal and decided that the two single-sex programs directly served Virginia’s reasserted purposes: single-gender education, and “achieving the results of an adversative method in a military environment.” See 44 F. 3d, at 1236, 1239. Inspecting the VMI and VWIL educational programs to determine whether they “afford[ed] to both genders benefits comparable in substance, [if] not in form and detail,” id., at 1240, the Court of Appeals concluded that Virginia had arranged for men and women opportunities “sufficiently comparable” to survive equal protection evaluation, id., at 1240-1241. The United States challenges this “remedial” ruling as pervasively misguided.

547*547 A

A remedial decree, this Court has said, must closely fit the constitutional violation; it must be shaped to place persons unconstitutionally denied an opportunity or advantage in “the position they would have occupied in the absence of [discrimination].” See Milliken v. Bradley, 433 U. S. 267, 280 (1977) (internal quotation marks omitted). The constitutional violation in this suit is the categorical exclusion of women from an extraordinary educational opportunity afforded men. A proper remedy for an unconstitutional exclusion, we have explained, aims to “eliminate [so far as possible] the discriminatory effects of the past” and to “bar like discrimination in the future.” Louisiana v. United States, 380 U. S. 145, 154 (1965).

Virginia chose not to eliminate, but to leave untouched, VMI’s exclusionary policy. For women only, however, Virginia proposed a separate program, different in kind from VMI and unequal in tangible and intangible facilities.[17] Having violated the Constitution’s equal protection requirement, Virginia was obliged to show that its remedial proposal “directly address[ed] and relate[d] to” the violation, see Milliken, 433 U. S., at 282, i. e., the equal protection denied to women ready, willing, and able to benefit from educational 548*548 opportunities of the kind VMI offers. Virginia described VWIL as a “parallel program,” and asserted that VWIL shares VMI’s mission of producing “citizen-soldiers” and VMI’s goals of providing “education, military training, mental and physical discipline, character . . . and leadership development.” Brief for Respondents 24 (internal quotation marks omitted). If the VWIL program could not “eliminate the discriminatory effects of the past,” could it at least “bar like discrimination in the future”? See Louisiana, 380 U. S., at 154. A comparison of the programs said to be “parallel” informs our answer. In exposing the character of, and differences in, the VMI and VWIL programs, we recapitulate facts earlier presented. See supra, at 520-523, 526-527.

VWIL affords women no opportunity to experience the rigorous military training for which VMI is famed. See 766 F. Supp., at 1413-1414 (“No other school in Virginia or in the United States, public or private, offers the same kind of rigorous military training as is available at VMI.”); id. , at 1421 (VMI “is known to be the most challenging military school in the United States”). Instead, the VWIL program “deemphasize[s]” military education, 44 F. 3d, at 1234, and uses a “cooperative method” of education “which reinforces self-esteem,” 852 F. Supp., at 476.

VWIL students participate in ROTC and a “largely ceremonial” Virginia Corps of Cadets, see 44 F. 3d, at 1234, but Virginia deliberately did not make VWIL a military institute. The VWIL House is not a military-style residence and VWIL students need not live together throughout the 4-year program, eat meals together, or wear uniforms during the schoolday. See 852 F. Supp., at 477, 495. VWIL students thus do not experience the “barracks” life “crucial to the VMI experience,” the spartan living arrangements designed to foster an “egalitarian ethic.” See 766 F. Supp., at 1423-1424. “[T]he most important aspects of the VMI educational experience occur in the barracks,” the District Court 549*549 found, id., at 1423, yet Virginia deemed that core experience nonessential, indeed inappropriate, for training its female citizen-soldiers.

VWIL students receive their “leadership training” in seminars, externships, and speaker series, see 852 F. Supp., at 477, episodes and encounters lacking the “[p]hysical rigor, mental stress, . .. minute regulation of behavior, and indoctrination in desirable values” made hallmarks of VMI’s citizen-soldier training, see 766 F. Supp., at 1421.[18] Kept away from the pressures, hazards, and psychological bonding characteristic of VMI’s adversative training, see id., at 1422, VWIL students will not know the “feeling of tremendous accomplishment” commonly experienced by VMI’s successful cadets, id. , at 1426.

Virginia maintains that these methodological differences are “justified pedagogically,” based on “important differences between men and women in learning and developmental needs,” “psychological and sociological differences” Virginia describes as “real” and “not stereotypes.” Brief for Respondents 28 (internal quotation marks omitted). The Task Force charged with developing the leadership program for women, drawn from the staff and faculty at Mary Baldwin College, “determined that a military model and, especially VMI’s adversative method, would be wholly inappropriate for educating and training most women. ” 852 F. Supp., at 476 (emphasis added). See also 44 F. 3d, at 1233-1234 (noting Task Force conclusion that, while “some women would be suited to and interested in [a VMI-style experience],” VMI’s adversative method “would not be effective for women as a group ” (emphasis added)). The Commonwealth 550*550 embraced the Task Force view, as did expert witnesses who testified for Virginia. See 852 F. Supp., at 480-481.

As earlier stated, see supra, at 541-542, generalizations about “the way women are,” estimates of what is appropriate for most women, no longer justify denying opportunity to women whose talent and capacity place them outside the average description. Notably, Virginia never asserted that VMI’s method of education suits most men. It is also revealing that Virginia accounted for its failure to make the VWIL experience “the entirely militaristic experience of VMI” on the ground that VWIL “is planned for women who do not necessarily expect to pursue military careers.” 852 F. Supp., at 478. By that reasoning, VMI’s “entirely militaristic” program would be inappropriate for men in general or as a group, for “[o]nly about 15% of VMI cadets enter career military service.” See 766 F. Supp., at 1432.

In contrast to the generalizations about women on which Virginia rests, we note again these dispositive realities: VMI’s “implementing methodology” is not “inherently unsuitable to women,” 976 F. 2d, at 899; “some women . . . do well under [the] adversative model,” 766 F. Supp., at 1434 (internal quotation marks omitted); “some women, at least, would want to attend [VMI] if they had the opportunity,” id., at 1414; “some women are capable of all of the individual activities required of VMI cadets,” id., at 1412, and “can meet the physical standards [VMI] now impose[s] on men,” 976 F. 2d, at 896. It is on behalf of these women that the United States has instituted this suit, and it is for them that a remedy must be crafted,[19] a remedy that will end their 551*551 exclusion from a state-supplied educational opportunity for which they are fit, a decree that will “bar like discrimination in the future.” Louisiana, 380 U. S., at 154.


In myriad respects other than military training, VWIL does not qualify as VMI’s equal. VWIL’s student body, faculty, course offerings, and facilities hardly match VMI’s. Nor can the VWIL graduate anticipate the benefits associated with VMI’s 157-year history, the school’s prestige, and its influential alumni network.

Mary Baldwin College, whose degree VWIL students will gain, enrolls first-year women with an average combined SAT score about 100 points lower than the average score for VMI freshmen. 852 F. Supp., at 501. The Mary Baldwin faculty holds “significantly fewer Ph. D.’s,” id. , at 502, and receives substantially lower salaries, see Tr. 158 (testimony of James Lott, Dean of Mary Baldwin College), than the faculty at VMI.

Mary Baldwin does not offer a VWIL student the range of curricular choices available to a VMI cadet. VMI awards baccalaureate degrees in liberal arts, biology, chemistry, civil engineering, electrical and computer engineering, and mechanical engineering. See 852 F. Supp., at 503; Virginia Military Institute: More than an Education 11 (Govt. exh. 75, 552*552 lodged with Clerk of this Court). VWIL students attend a school that “does not have a math and science focus,” 852 F. Supp., at 503; they cannot take at Mary Baldwin any courses in engineering or the advanced math and physics courses VMI offers, see id., at 477.

For physical training, Mary Baldwin has “two multipurpose fields” and “[o]ne gymnasium.” Id. , at 503. VMI has “an NCAA competition level indoor track and field facility; a number of multi-purpose fields; baseball, soccer and lacrosse fields; an obstacle course; large boxing, wrestling and martial arts facilities; an 11-laps-to-the-mile indoor running course; an indoor pool; indoor and outdoor rifle ranges; and a football stadium that also contains a practice field and outdoor track.” Ibid.

Although Virginia has represented that it will provide equal financial support for in-state VWIL students and VMI cadets, id. , at 483, and the VMI Foundation has agreed to endow VWIL with $5.4625 million, id. , at 499, the difference between the two schools’ financial reserves is pronounced. Mary Baldwin’s endowment, currently about $19 million, will gain an additional $35 million based on future commitments; VMI’s current endowment, $131 million—the largest public college per-student endowment in the Nation—will gain $220 million. Id. , at 503.

The VWIL student does not graduate with the advantage of a VMI degree. Her diploma does not unite her with the legions of VMI “graduates [who] have distinguished themselves” in military and civilian life. See 976 F. 2d, at 892-893. “[VMI] alumni are exceptionally close to the school,” and that closeness accounts, in part, for VMI’s success in attracting applicants. See 766 F. Supp., at 1421. A VWIL graduate cannot assume that the “network of business owners, corporations, VMI graduates and non-graduate employers . . . interested in hiring VMI graduates,” 852 F. Supp., at 499, will be equally responsive to her search for employment, 553*553 see 44 F. 3d, at 1250 (Phillips, J., dissenting) (“the powerful political and economic ties of the VMI alumni network cannot be expected to open” for graduates of the fledgling VWIL program).

Virginia, in sum, while maintaining VMI for men only, has failed to provide any “comparable single-gender women’s institution.” Id. , at 1241. Instead, the Commonwealth has created a VWIL program fairly appraised as a “pale shadow” of VMI in terms of the range of curricular choices and faculty stature, funding, prestige, alumni support and influence. See id., at 1250 (Phillips, J., dissenting).

Virginia’s VWIL solution is reminiscent of the remedy Texas proposed 50 years ago, in response to a state trial court’s 1946 ruling that, given the equal protection guarantee, African-Americans could not be denied a legal education at a state facility. See Sweatt v. Painter, 339 U. S. 629 (1950). Reluctant to admit African-Americans to its flagship University of Texas Law School, the State set up a separate school for Heman Sweatt and other black law students. Id. , at 632. As originally opened, the new school had no independent faculty or library, and it lacked accreditation. Id. , at 633. Nevertheless, the state trial and appellate courts were satisfied that the new school offered Sweatt opportunities for the study of law “substantially equivalent to those offered by the State to white students at the University of Texas.” Id. , at 632 (internal quotation marks omitted).

Before this Court considered the case, the new school had gained “a faculty of five full-time professors; a student body of 23; a library of some 16,500 volumes serviced by a full-time staff; a practice court and legal aid association; and one alumnus who ha[d] become a member of the Texas Bar.” Id. , at 633. This Court contrasted resources at the new school with those at the school from which Sweatt had been excluded. The University of Texas Law School had a full-time faculty of 16, a student body of 850, a library containing over 554*554 65,000 volumes, scholarship funds, a law review, and moot court facilities. Id. , at 632-633.

More important than the tangible features, the Court emphasized, are “those qualities which are incapable of objective measurement but which make for greatness” in a school, including “reputation of the faculty, experience of the administration, position and influence of the alumni, standing in the community, traditions and prestige.” Id. , at 634. Facing the marked differences reported in the Sweatt opinion, the Court unanimously ruled that Texas had not shown “substantial equality in the [separate] educational opportunities” the State offered. Id. , at 633. Accordingly, the Court held, the Equal Protection Clause required Texas to admit AfricanAmericans to the University of Texas Law School. Id. , at 636. In line with Sweatt, we rule here that Virginia has not shown substantial equality in the separate educational opportunities the Commonwealth supports at VWIL and VMI.


When Virginia tendered its VWIL plan, the Fourth Circuit did not inquire whether the proposed remedy, approved by the District Court, placed women denied the VMI advantage in “the position they would have occupied in the absence of [discrimination].” Milliken, 433 U. S., at 280 (internal quotation marks omitted). Instead, the Court of Appeals considered whether the Commonwealth could provide, with fidelity to the equal protection principle, separate and unequal educational programs for men and women.

The Fourth Circuit acknowledged that “the VWIL degree from Mary Baldwin College lacks the historical benefit and prestige of a degree from VMI.” 44 F. 3d, at 1241. The Court of Appeals further observed that VMI is “an ongoing and successful institution with a long history,” and there remains no “comparable single-gender women’s institution.” Ibid. Nevertheless, the appeals court declared the substantially different and significantly unequal VWIL program satisfactory. 555*555 The court reached that result by revising the applicable standard of review. The Fourth Circuit displaced the standard developed in our precedent, see supra, at 532-534, and substituted a standard of its own invention.

We have earlier described the deferential review in which the Court of Appeals engaged, see supra, at 528-529, a brand of review inconsistent with the more exacting standard our precedent requires, see supra, at 532-534. Quoting in part from Mississippi Univ. for Women, the Court of Appeals candidly described its own analysis as one capable of checking a legislative purpose ranked as “pernicious,” but generally according “deference to [the] legislative will.” 44 F. 3d, at 1235, 1236. Recognizing that it had extracted from our decisions a test yielding “little or no scrutiny of the effect of a classification directed at [single-gender education],” the Court of Appeals devised another test, a “substantive comparability” inquiry, id., at 1237, and proceeded to find that new test satisfied, id. , at 1241.

The Fourth Circuit plainly erred in exposing Virginia’s VWIL plan to a deferential analysis, for “all gender-based classifications today” warrant “heightened scrutiny.” See J. E. B., 511 U. S., at 136. Valuable as VWIL may prove for students who seek the program offered, Virginia’s remedy affords no cure at all for the opportunities and advantages withheld from women who want a VMI education and can make the grade. See supra, at 549-554.[20] In sum, Virginia’s 556*556 remedy does not match the constitutional violation; the Commonwealth has shown no “exceedingly persuasive justification” for withholding from women qualified for the experience premier training of the kind VMI affords.


A generation ago, “the authorities controlling Virginia higher education,” despite long established tradition, agreed “to innovate and favorably entertain[ed] the [then] relatively new idea that there must be no discrimination by sex in offering educational opportunity.” Kirstein, 309 F. Supp., at 186. Commencing in 1970, Virginia opened to women “educational opportunities at the Charlottesville campus that [were] not afforded in other [state-operated] institutions.” Id., at 187; see supra, at 538. A federal court approved the Commonwealth’s innovation, emphasizing that the University of Virginia “offer[ed] courses of instruction . . . not available elsewhere.” 309 F. Supp., at 187. The court further noted: “[T]here exists at Charlottesville a `prestige’ factor 557*557 [not paralleled in] other Virginia educational institutions.” Ibid.

VMI, too, offers an educational opportunity no other Virginia institution provides, and the school’s “prestige”—associated with its success in developing “citizen-soldiers”—is unequaled. Virginia has closed this facility to its daughters and, instead, has devised for them a “parallel program,” with a faculty less impressively credentialed and less well paid, more limited course offerings, fewer opportunities for military training and for scientific specialization. Cf. Sweatt, 339 U. S., at 633. VMI, beyond question, “possesses to a far greater degree” than the VWIL program “those qualities which are incapable of objective measurement but which make for greatness in a . . . school,” including “position and influence of the alumni, standing in the community, traditions and prestige.” Id., at 634. Women seeking and fit for a VMI-quality education cannot be offered anything less, under the Commonwealth’s obligation to afford them genuinely equal protection.

A prime part of the history of our Constitution, historian Richard Morris recounted, is the story of the extension of constitutional rights and protections to people once ignored or excluded.[21] VMI’s story continued as our comprehension of “We the People” expanded. See supra, at 532, n. 6. 558*558 There is no reason to believe that the admission of women capable of all the activities required of VMI cadets would destroy the Institute rather than enhance its capacity to serve the “more perfect Union.”

* * *

For the reasons stated, the initial judgment of the Court of Appeals, 976 F. 2d 890 (CA4 1992), is affirmed, the final judgment of the Court of Appeals, 44 F. 3d 1229 (CA4 1995), is reversed, and the case is remanded for further proceedings consistent with this opinion.

It is so ordered.

Justice Thomas took no part in the consideration or decision of these cases.

Chief Justice Rehnquist, concurring in the judgment.

The Court holds first that Virginia violates the Equal Protection Clause by maintaining the Virginia Military Institute’s (VMI’s) all-male admissions policy, and second that establishing the Virginia Women’s Institute for Leadership (VWIL) program does not remedy that violation. While I agree with these conclusions, I disagree with the Court’s analysis and so I write separately.


Two decades ago in Craig v. Boren, 429 U. S. 190, 197 (1976), we announced that “[t]o withstand constitutional challenge, . . . classifications by gender must serve important governmental objectives and must be substantially related to achievement of those objectives.” We have adhered to that standard of scrutiny ever since. See Califano v. Goldfarb, 430 U. S. 199, 210-211 (1977); Califano v. Webster, 430 U. S. 313, 316-317 (1977); Orr v. Orr, 440 U. S. 268, 279 (1979); Caban v. Mohammed, 441 U. S. 380, 388 (1979); Davis v. Passman, 442 U. S. 228, 234-235, 235, n. 9 (1979); Personnel Administrator of Mass. v. Feeney, 442 U. S. 256, 273 (1979); 559*559 Califano v. Westcott, 443 U. S. 76, 85 (1979); Wengler v. Druggists Mut. Ins. Co., 446 U. S. 142, 150 (1980); Kirchberg v. Feenstra, 450 U. S. 455, 459-460 (1981); Michael M. v. Superior Court, Sonoma Cty., 450 U. S. 464, 469 (1981); Mississippi Univ. for Women v. Hogan, 458 U. S. 718, 724 (1982); Heckler v. Mathews, 465 U. S. 728, 744 (1984); J. E. B. v. Alabama ex rel. T. B., 511 U. S. 127, 137, n. 6 (1994). While the majority adheres to this test today, ante, at 524, 533, it also says that the Commonwealth must demonstrate an “`exceedingly persuasive justification’ ” to support a gender-based classification. See ante, at 524, 529, 530, 531, 533, 534, 545, 546, 556. It is unfortunate that the Court thereby introduces an element of uncertainty respecting the appropriate test.

While terms like “important governmental objective” and “substantially related” are hardly models of precision, they have more content and specificity than does the phrase “exceedingly persuasive justification.” That phrase is best confined, as it was first used, as an observation on the difficulty of meeting the applicable test, not as a formulation of the test itself. See, e. g., Feeney, supra, at 273 (“[T]hese precedents dictate that any state law overtly or covertly designed to prefer males over females in public employment require an exceedingly persuasive justification”). To avoid introducing potential confusion, I would have adhered more closely to our traditional, “firmly established,” Hogan, supra, at 723; Heckler, supra, at 744, standard that a gender-based classification “must bear a close and substantial relationship to important governmental objectives.” Feeney, supra, at 273.

Our cases dealing with gender discrimination also require that the proffered purpose for the challenged law be the actual purpose. See ante, at 533, 535-536. It is on this ground that the Court rejects the first of two justifications Virginia offers for VMI’s single-sex admissions policy, namely, the goal of diversity among its public educational institutions. While I ultimately agree that the Commonwealth 560*560 has not carried the day with this justification, I disagree with the Court’s method of analyzing the issue.

VMI was founded in 1839, and, as the Court notes, ante, at 536-537, admission was limited to men because under the then-prevailing view men, not women, were destined for higher education. However misguided this point of view may be by present-day standards, it surely was not unconstitutional in 1839. The adoption of the Fourteenth Amendment, with its Equal Protection Clause, was nearly 30 years in the future. The interpretation of the Equal Protection Clause to require heightened scrutiny for gender discrimination was yet another century away.

Long after the adoption of the Fourteenth Amendment, and well into this century, legal distinctions between men and women were thought to raise no question under the Equal Protection Clause. The Court refers to our decision in Goesaert v. Cleary, 335 U. S. 464 (1948). Likewise representing that now abandoned view was Hoyt v. Florida, 368 U. S. 57 (1961), where the Court upheld a Florida system of jury selection in which men were automatically placed on jury lists, but women were placed there only if they expressed an affirmative desire to serve. The Court noted that despite advances in women’s opportunities, the “woman is still regarded as the center of home and family life.” Id. , at 62.

Then, in 1971, we decided Reed v. Reed, 404 U. S. 71, which the Court correctly refers to as a seminal case. But its facts have nothing to do with admissions to any sort of educational institution. An Idaho statute governing the administration of estates and probate preferred men to women if the other statutory qualifications were equal. The statute’s purpose, according to the Idaho Supreme Court, was to avoid hearings to determine who was better qualified as between a man and a woman both applying for letters of administration. This Court held that such a rule violated the Fourteenth Amendment because “a mandatory preference to members of either 561*561 sex over members of the other, merely to accomplish the elimination of hearings,” was an “arbitrary legislative choice forbidden by the Equal Protection Clause.” Id. , at 76. The brief opinion in Reed made no mention of either Goesaert or Hoyt.

Even at the time of our decision in Reed v. Reed, therefore, Virginia and VMI were scarcely on notice that its holding would be extended across the constitutional board. They were entitled to believe that “one swallow doesn’t make a summer” and await further developments. Those developments were 11 years in coming. In Mississippi Univ. for Women v. Hogan, supra, a case actually involving a singlesex admissions policy in higher education, the Court held that the exclusion of men from a nursing program violated the Equal Protection Clause. This holding did place Virginia on notice that VMI’s men-only admissions policy was open to serious question.

The VMI Board of Visitors, in response, appointed a Mission Study Committee to examine “the legality and wisdom of VMI’s single-sex policy in light of” Hogan. 766 F. Supp. 1407, 1427 (WD Va. 1991). But the committee ended up cryptically recommending against changing VMI’s status as a single-sex college. After three years of study, the committee found “`no information’ ” that would warrant a change in VMI’s status. Id. , at 1429. Even the District Court, ultimately sympathetic to VMI’s position, found that “[t]he Report provided very little indication of how [its] conclusion was reached” and that “[t]he one and one-half pages in the committee’s final report devoted to analyzing the information it obtained primarily focuses on anticipated difficulties in attracting females to VMI.” Ibid. The reasons given in the report for not changing the policy were the changes that admission of women to VMI would require, and the likely effect of those changes on the institution. That VMI would have to change is simply not helpful in addressing the constitutionality of the status after Hogan.

562*562 Before this Court, Virginia has sought to justify VMI’s single-sex admissions policy primarily on the basis that diversity in education is desirable, and that while most of the public institutions of higher learning in the Commonwealth are coeducational, there should also be room for single-sex institutions. I agree with the Court that there is scant evidence in the record that this was the real reason that Virginia decided to maintain VMI as men only.[*] But, unlike the majority, I would consider only evidence that postdates our decision in Hogan, and would draw no negative inferences from the Commonwealth’s actions before that time. I think that after Hogan, the Commonwealth was entitled to reconsider its policy with respect to VMI, and not to have earlier justifications, or lack thereof, held against it.

Even if diversity in educational opportunity were the Commonwealth’s actual objective, the Commonwealth’s position would still be problematic. The difficulty with its position is that the diversity benefited only one sex; there was single-sex public education available for men at VMI, but no corresponding single-sex public education available for women. When Hogan placed Virginia on notice that 563*563 VMI’s admissions policy possibly was unconstitutional, VMI could have dealt with the problem by admitting women; but its governing body felt strongly that the admission of women would have seriously harmed the institution’s educational approach. Was there something else the Commonwealth could have done to avoid an equal protection violation? Since the Commonwealth did nothing, we do not have to definitively answer that question.

I do not think, however, that the Commonwealth’s options were as limited as the majority may imply. The Court cites, without expressly approving it, a statement from the opinion of the dissenting judge in the Court of Appeals, to the effect that the Commonwealth could have “simultaneously opened single-gender undergraduate institutions having substantially comparable curricular and extra-curricular programs, funding, physical plant, administration and support services, and faculty and library resources.” Ante, at 529-530 (internal quotation marks omitted). If this statement is thought to exclude other possibilities, it is too stringent a requirement. VMI had been in operation for over a century and a half, and had an established, successful, and devoted group of alumni. No legislative wand could instantly call into existence a similar institution for women; and it would be a tremendous loss to scrap VMI’s history and tradition. In the words of Grover Cleveland’s second inaugural address, the Commonwealth faced a condition, not a theory. And it was a condition that had been brought about, not through defiance of decisions construing gender bias under the Equal Protection Clause, but, until the decision in Hogan, a condition that had not appeared to offend the Constitution. Had Virginia made a genuine effort to devote comparable public resources to a facility for women, and followed through on such a plan, it might well have avoided an equal protection violation. I do not believe the Commonwealth was faced with the stark choice of either admitting women to VMI, on the 564*564 one hand, or abandoning VMI and starting from scratch for both men and women, on the other.

But, as I have noted, neither the governing board of VMI nor the Commonwealth took any action after 1982. If diversity in the form of single-sex, as well as coeducational, institutions of higher learning were to be available to Virginians, that diversity had to be available to women as well as to men.

The dissent criticizes me for “disregarding the four allwomen’s private colleges in Virginia (generously assisted by public funds).” Post, at 595. The private women’s colleges are treated by the Commonwealth exactly as all other private schools are treated, which includes the provision of tuition-assistance grants to Virginia residents. Virginia gives no special support to the women’s single-sex education. But obviously, the same is not true for men’s education. Had the Commonwealth provided the kind of support for the private women’s schools that it provides for VMI, this may have been a very different case. For in so doing, the Commonwealth would have demonstrated that its interest in providing a single-sex education for men was to some measure matched by an interest in providing the same opportunity for women.

Virginia offers a second justification for the single-sex admissions policy: maintenance of the adversative method. I agree with the Court that this justification does not serve an important governmental objective. A State does not have substantial interest in the adversative methodology unless it is pedagogically beneficial. While considerable evidence shows that a single-sex education is pedagogically beneficial for some students, see 766 F. Supp., at 1414, and hence a State may have a valid interest in promoting that methodology, there is no similar evidence in the record that an adversative method is pedagogically beneficial or is any more likely to produce character traits than other methodologies.

565*565 II

The Court defines the constitutional violation in these cases as “the categorical exclusion of women from an extraordinary educational opportunity afforded to men.” Ante, at 547. By defining the violation in this way, and by emphasizing that a remedy for a constitutional violation must place the victims ofdiscrimination in “`the position they would have occupied in the absence of [discrimination],’ ” ibid. , the Court necessarily implies that the only adequate remedy would be the admission of women to the allmale institution. As the foregoing discussion suggests, I would not define the violation in this way; it is not the “exclusion of women” that violates the Equal Protection Clause, but the maintenance of an all-men school without providing any—much less a comparable—institution for women.

Accordingly, the remedy should not necessarily require either the admission of women to VMI or the creation of a VMI clone for women. An adequate remedy in my opinion might be a demonstration by Virginia that its interest in educating men in a single-sex environment is matched by its interest in educating women in a single-sex institution. To demonstrate such, the Commonwealth does not need to create two institutions with the same number of faculty Ph. D.’s, similar SAT scores, or comparable athletic fields. See ante, at 551-552. Nor would it necessarily require that the women’s institution offer the same curriculum as the men’s; one could be strong in computer science, the other could be strong in liberal arts. It would be a sufficient remedy, I think, if the two institutions offered the same quality of education and were of the same overall caliber.

If a State decides to create single-sex programs, the State would, I expect, consider the public’s interest and demand in designing curricula. And rightfully so. But the State should avoid assuming demand based on stereotypes; it must not assume a priori, without evidence, that there would be 566*566 no interest in a women’s school of civil engineering, or in a men’s school of nursing.

In the end, the women’s institution Virginia proposes, VWIL, fails as a remedy, because it is distinctly inferior to the existing men’s institution and will continue to be for the foreseeable future. VWIL simply is not, in any sense, the institution that VMI is. In particular, VWIL is a program appended to a private college, not a self-standing institution; and VWIL is substantially underfunded as compared to VMI. I therefore ultimately agree with the Court that Virginia has not provided an adequate remedy.

Justice Scalia, dissenting.

Today the Court shuts down an institution that has served the people of the Commonwealth of Virginia with pride and distinction for over a century and a half. To achieve that desired result, it rejects (contrary to our established practice) the factual findings of two courts below, sweeps aside the precedents of this Court, and ignores the history of our people. As to facts: It explicitly rejects the finding that there exist “gender-based developmental differences” supporting Virginia’s restriction of the “adversative” method to only a men’s institution, and the finding that the all-male composition of the Virginia Military Institute (VMI) is essential to that institution’s character. As to precedent: It drastically revises our established standards for reviewing sex-based classifications. And as to history: It counts for nothing the long tradition, enduring down to the present, of men’s military colleges supported by both States and the Federal Government.

Much of the Court’s opinion is devoted to deprecating the closed-mindedness of our forebears with regard to women’s education, and even with regard to the treatment of women in areas that have nothing to do with education. Closedminded they were—as every age is, including our own, with regard to matters it cannot guess, because it simply does not 567*567 consider them debatable. The virtue of a democratic system with a First Amendment is that it readily enables the people, over time, to be persuaded that what they took for granted is not so, and to change their laws accordingly. That system is destroyed if the smug assurances of each age are removed from the democratic process and written into the Constitution. So to counterbalance the Court’s criticism of our ancestors, let me say a word in their praise: They left us free to change. The same cannot be said of this most illiberal Court, which has embarked on a course of inscribing one after another of the current preferences of the society (and in some cases only the countermajoritarian preferences of the society’s law-trained elite) into our Basic Law. Today it enshrines the notion that no substantial educational value is to be served by an all-men’s military academy—so that the decision by the people of Virginia to maintain such an institution denies equal protection to women who cannot attend that institution but can attend others. Since it is entirely clear that the Constitution of the United States—the old one—takes no sides in this educational debate, I dissent.


I shall devote most of my analysis to evaluating the Court’s opinion on the basis of our current equal protection jurisprudence, which regards this Court as free to evaluate everything under the sun by applying one of three tests: “rational basis” scrutiny, intermediate scrutiny, or strict scrutiny. These tests are no more scientific than their names suggest, and a further element of randomness is added by the fact that it is largely up to us which test will be applied in each case. Strict scrutiny, we have said, is reserved for state “classifications based on race or national origin and classifications affecting fundamental rights,” Clark v. Jeter, 486 U. S. 456, 461 (1988) (citation omitted). It is my position that the term “fundamental rights” should be limited to “interest[s] traditionally protected by our society,” Michael H. 568*568 v. Gerald D., 491 U. S. 110, 122 (1989) (plurality opinion of Scalia, J.); but the Court has not accepted that view, so that strict scrutiny will be applied to the deprivation of whatever sort of right we consider “fundamental.” We have no established criterion for “intermediate scrutiny” either, but essentially apply it when it seems like a good idea to load the dice. So far it has been applied to content-neutral restrictions that place an incidental burden on speech, to disabilities attendant to illegitimacy, and to discrimination on the basis of sex. See, e. g., Turner Broadcasting System, Inc. v. FCC, 512 U. S. 622, 662 (1994); Mills v. Habluetzel, 456 U. S. 91, 98-99 (1982); Craig v. Boren, 429 U. S. 190, 197 (1976).

I have no problem with a system of abstract tests such as rational basis, intermediate, and strict scrutiny (though I think we can do better than applying strict scrutiny and intermediate scrutiny whenever we feel like it). Such formulas are essential to evaluating whether the new restrictions that a changing society constantly imposes upon private conduct comport with that “equal protection” our society has always accorded in the past. But in my view the function of this Court is to preserve our society’s values regarding (among other things) equal protection, not to revise them; to prevent backsliding from the degree of restriction the Constitution imposed upon democratic government, not to prescribe, on our own authority, progressively higher degrees. For that reason it is my view that, whatever abstract tests we may choose to devise, they cannot supersede—and indeed ought to be crafted so as to reflect —those constant and unbroken national traditions that embody the people’s understanding of ambiguous constitutional texts. More specifically, it is my view that “when a practice not expressly prohibited by the text of the Bill of Rights bears the endorsement of a long tradition of open, widespread, and unchallenged use that dates back to the beginning of the Republic, we have no proper basis for striking it down.” Rutan v. Republican Party of Ill., 497 U. S. 62, 95 (1990) (Scalia, J., 569*569 dissenting). The same applies, mutatis mutandis, to a practice asserted to be in violation of the post-Civil War Fourteenth Amendment. See, e. g., Burnham v. Superior Court of Cal., County of Marin, 495 U. S. 604 (1990) (plurality opinion of Scalia, J.) (Due Process Clause); J. E. B. v. Alabama ex rel. T. B., 511 U. S. 127, 156-163 (1994) (Scalia, J., dissenting) (Equal Protection Clause); Planned Parenthood of Southeastern Pa. v. Casey, 505 U. S. 833, 979-984, 1000-1001 (1992) (Scalia, J., dissenting) (various alleged “penumbras”).

The all-male constitution of VMI comes squarely within such a governing tradition. Founded by the Commonwealth of Virginia in 1839 and continuously maintained by it since, VMI has always admitted only men. And in that regard it has not been unusual. For almost all of VMI’s more than a century and a half of existence, its single-sex status reflected the uniform practice for government-supported military colleges. Another famous Southern institution, The Citadel, has existed as a state-funded school of South Carolina since 1842. And all the federal military colleges—West Point, the Naval Academy at Annapolis, and even the Air Force Academy, which was not established until 1954—admitted only males for most of their history. Their admission of women in 1976 (upon which the Court today relies, see ante, at 544-545, nn. 13, 15) came not by court decree, but because the people, through their elected representatives, decreed a change. See, e. g., § 803(a), 89 Stat. 537, note following 10 U. S. C. § 4342. In other words, the tradition of having government-funded military schools for men is as well rooted in the traditions of this country as the tradition of sending only men into military combat. The people may decide to change the one tradition, like the other, through democratic processes; but the assertion that either tradition has been unconstitutional through the centuries is not law, but politics-smuggled-into-law.

And the same applies, more broadly, to single-sex education in general, which, as I shall discuss, is threatened by 570*570 today’s decision with the cutoff of all state and federal support. Government-run non military educational institutions for the two sexes have until very recently also been part of our national tradition. “[It is] [c]oeducation, historically, [that] is a novel educational theory. From grade school through high school, college, and graduate and professional training, much of the Nation’s population during much of our history has been educated in sexually segregated classrooms.” Mississippi Univ. for Women v. Hogan, 458 U. S. 718, 736 (1982) (Powell, J.,dissenting); see id., at 736-739. These traditions may of course be changed by the democratic decisions of the people, as they largely have been.

Today, however, change is forced upon Virginia, and reversion to single-sex education is prohibited nationwide, not by democratic processes but by order of this Court. Even while bemoaning the sorry, bygone days of “fixed notions” concerning women’s education, see ante, at 536-537, and n. 10, 537-539, 542-544, the Court favors current notions so fixedly that it is willing to write them into the Constitution of the United States by application of custom-built “tests.” This is not the interpretation of a Constitution, but the creation of one.


To reject the Court’s disposition today, however, it is not necessary to accept my view that the Court’s made-up tests cannot displace longstanding national traditions as the primary determinant of what the Constitution means. It is only necessary to apply honestly the test the Court has been applying to sex-based classifications for the past two decades. It is well settled, as Justice O’Connor stated some time ago for a unanimous Court, that we evaluate a statutory classification based on sex under a standard that lies “[b]etween th[e] extremes of rational basis review and strict scrutiny.” Clark v. Jeter, 486 U. S., at 461. We have denominated this standard “intermediate scrutiny” and under it have inquired whether the statutory classification is “substantially 571*571 related to an important governmental objective.” Ibid. See, e. g., Heckler v. Mathews, 465 U. S. 728, 744 (1984); Wengler v. Druggists Mut. Ins. Co., 446 U. S. 142, 150 (1980); Craig v. Boren, 429 U. S., at 197.

Before I proceed to apply this standard to VMI, I must comment upon the manner in which the Court avoids doing so. Notwithstanding our above-described precedents and their “`firmly established principles,’ ” Heckler, supra, at 744 (quoting Hogan, supra, at 723), the United States urged us to hold in this litigation “that strict scrutiny is the correct constitutional standard for evaluating classifications that deny opportunities to individuals based on their sex.” Brief for United States in No. 94-2107, p. 16. (This was in flat contradiction of the Government’s position below, which was, in its own words, to “stat[e] unequivocally that the appropriate standard in this case is `intermediate scrutiny.’ ” 2 Record, Doc. No. 88, p. 3 (emphasis added).) The Court, while making no reference to the Government’s argument, effectively accepts it.

Although the Court in two places recites the test as stated in Hogan, see ante, at 524, 532-533, which asks whether the State has demonstrated “that the classification serves important governmental objectives and that the discriminatory means employed are substantially related to the achievement of those objectives,” 458 U. S., at 724 (internal quotation marks omitted), the Court never answers the question presented in anything resembling that form. When it engages in analysis, the Court instead prefers the phrase “exceedingly persuasive justification” from Hogan. The Court’s nine invocations of that phrase, see ante, at 524, 529, 530, 531, 533, 534, 545, 546, 556, and even its fanciful description of that imponderable as “the core instruction” of the Court’s decisions in J. E. B. v. Alabama ex rel. T. B., supra, and Hogan, supra, see ante, at 531, would be unobjectionable if the Court acknowledged that whether a “justification” is “exceedingly persuasive” must be assessed by asking 572*572 “[whether] the classification serves important governmental objectives and [whether] the discriminatory means employed are substantially related to the achievement of those objectives.” Instead, however, the Court proceeds to interpret “exceedingly persuasive justification” in a fashion that contradicts the reasoning of Hogan and our other precedents.

That is essential to the Court’s result, which can only be achieved by establishing that intermediate scrutiny is not survived if there are some women interested in attending VMI, capable of undertaking its activities, and able to meet its physical demands. Thus, the Court summarizes its holding as follows:

“In contrast to the generalizations about women on which Virginia rests, we note again these dispositive realities: VMI’s implementing methodology is not inher- ently unsuitable to women; some women do well under the adversative model; some women, at least, would want to attend VMI if they had the opportunity; some women are capable of all of the individual activities required of VMI cadets and can meet the physical standards VMI now imposes on men.” Ante, at 550 (internal quotation marks, citations, and punctuation omitted; emphasis added).

Similarly, the Court states that “[t]he Commonwealth’s justification for excluding all women from `citizen-soldier’ training for which some are qualified . . . cannot rank as `exceedingly persuasive’ . . . .” Ante, at 545.[1]

573*573 Only the amorphous “exceedingly persuasive justification” phrase, and not the standard elaboration of intermediate scrutiny, can be made to yield this conclusion that VMI’s single-sex composition is unconstitutional because there exist several women (or, one would have to conclude under the Court’s reasoning, a single woman) willing and able to undertake VMI’s program. Intermediate scrutiny has never required a least-restrictive-means analysis, but only a “substantial relation” between the classification and the state interests that it serves. Thus, in Califano v. Webster, 430 U. S. 313 (1977) (per curiam), we upheld a congressional statute that provided higher Social Security benefits for women than for men. We reasoned that “women . . . as such have been unfairly hindered from earning as much as men,” but we did not require proof that each woman so benefited had suffered discrimination or that each disadvantaged man had not; it was sufficient that even under the former congressional scheme “women on the average received lower retirement benefits than men.” Id., at 318, and n. 5 (emphasis added). The reasoning in our other intermediate-scrutiny cases has similarly required only a substantial relation between end and means, not a perfect fit. In Rostker v. Goldberg, 453 U. S. 57 (1981), we held that selective-service registration could constitutionally exclude women, because even “assuming that a small number of women could be drafted for noncombat roles, Congress simply did not consider it worth the added burdens of including women in draft and registration plans.” Id., at 81. In Metro Broadcasting, Inc. v. FCC, 497 U. S. 547, 579, 582-583 (1990), overruled on other grounds, Adarand Constructors, Inc. v. Peña, 515 U. S. 200, 227 (1995), we held that a classification need not be accurate “in every case” to survive intermediate scrutiny so long as, “in the aggregate,” it advances the underlying 574*574 objective. There issimply no support in our cases for the notion that a sex-based classification is invalid unless it relates to characteristics that hold true in every instance.

Not content to execute a de facto abandonment of the intermediate scrutiny that has been our standard for sex-based classifications for some two decades, the Court purports to reserve the question whether, even in principle, a higher standard (i. e., strict scrutiny) should apply. “The Court has,” it says, “thus far reserved most stringent judicial scrutiny for classifications based on race or national origin . . . ,” ante, at 532, n. 6 (emphasis added); and it describes our earlier cases as having done no more than decline to “equat[e] gender classifications, for all purposes, to classifications based on race or national origin,” ante, at 532 (emphasis added). The wonderful thing about these statements is that they are not actually false—just as it would not be actually false to say that “our cases have thus far reserved the `beyond a reasonable doubt’ standard of proof for criminal cases,” or that “we have not equated tort actions, for all purposes, to criminal prosecutions.” But the statements are misleading, insofar as they suggest that we have not already categorically held strict scrutiny to be inapplicable to sexbased classifications. See, e. g., Heckler v. Mathews, 465 U. S. 728 (1984) (upholding state action after applying only intermediate scrutiny); Michael M. v. Superior Court, Somoma Cty., 450 U. S. 464 (1981) (plurality and both concurring opinions) (same); Califano v. Webster, supra (per curiam) (same). And the statements are irresponsible, insofar as they are calculated to destabilize current law. Our task is to clarify the law—not to muddy the waters, and not to exact overcompliance by intimidation. The States and the Federal Government are entitled to know before they act the standard to which they will be held, rather than be compelled to guess about the outcome of Supreme Court peek-a-boo.

The Court’s intimations are particularly out of place because it is perfectly clear that, if the question of the applicable 575*575 standard of review for sex-based classifications were to be regarded as an appropriate subject for reconsideration, the stronger argument would be not for elevating the standard to strict scrutiny, but for reducing it to rational-basis review. The latter certainly has a firmer foundation in our past jurisprudence: Whereas no majority of the Court has ever applied strict scrutiny in a case involving sex-based classifications, we routinely applied rational-basis review until the 1970’s, see, e. g., Hoyt v. Florida, 368 U. S. 57 (1961); Goesaert v. Cleary, 335 U. S. 464 (1948). And of course normal, rational-basis review of sex-based classifications would be much more in accord with the genesis of heightened standards of judicial review, the famous footnote in United States v. Carolene Products Co., 304 U. S. 144 (1938), which said (intimatingly) that we did not have to inquire in the case at hand

“whether prejudice against discrete and insular minorities may be a special condition, which tends seriously to curtail the operation of those political processes ordinarily to be relied upon to protect minorities, and which may call for a correspondingly more searching judicial inquiry.” Id., at 152-153, n. 4.

It is hard to consider women a “discrete and insular minorit[y]” unable to employ the “political processes ordinarily to be relied upon,” when they constitute a majority of the electorate. And the suggestion that they are incapable of exerting that political power smacks of the same paternalism that the Court so roundly condemns. See, e. g., ante, at 536-537, 542-546 (and accompanying notes). Moreover, a long list of legislation proves the proposition false. See, e. g., Equal Pay Act of 1963, 29 U. S. C. § 206(d); Title VII of the Civil Rights Act of 1964, 42 U. S. C. § 2000e-2; Title IX of the Education Amendments of 1972, 20 U. S. C. § 1681; Women’s Business Ownership Act of 1988, Pub. L. 100-533, 102 Stat. 2689; 576*576 Violence Against Women Act of 1994, Pub. L. 103-322, Title IV, 108 Stat. 1902.


With this explanation of how the Court has succeeded in making its analysis seem orthodox—and indeed, if intimations are to be believed, even overly generous to VMI—I now proceed to describe how the analysis should have been conducted. The question to be answered, I repeat, is whether the exclusion of women from VMI is “substantially related to an important governmental objective.”


It is beyond question that Virginia has an important state interest in providing effective college education for itscitizens. That single-sex instruction is an approach substantially related to that interest should be evident enough from the long and continuing history in this country of men’s and women’s colleges. But beyond that, as the Court of Appeals here stated: “That single-gender education at the college level is beneficial to both sexes is a fact established in this case. ” 44 F. 3d 1229, 1238 (CA4 1995) (emphasis added).

The evidence establishing that fact was overwhelming— indeed, “virtually uncontradicted” in the words of the court that received the evidence, 766 F. Supp. 1407, 1415 (WD Va. 1991). As an initial matter, Virginia demonstrated at trial that “[a] substantial body of contemporary scholarship and research supports the proposition that, although males and females have significant areas of developmental overlap, they also have differing developmental needs that are deepseated.” Id., at 1434. While no one questioned that for many students a coeducational environment was nonetheless not inappropriate, that could not obscure the demonstrated benefits of single-sex colleges. For example, the District Court stated as follows:

“One empirical study in evidence, not questioned by any expert, demonstrates that single-sex colleges provide 577*577 better educational experiences than coeducational institutions. Students of both sexes become more academically involved, interact with faculty frequently, show larger increases in intellectual self-esteem and are more satisfied with practically all aspects of college experience (the sole exception is social life) compared with their counterparts in coeducational institutions. Attendance at an all-male college substantially increases the likelihood that a student will carry out career plans in law, business and college teaching, and also has a substantial positive effect on starting salaries in business. Women’s colleges increase the chances that those who attend will obtain positions of leadership, complete the baccalaureate degree, and aspire to higher degrees.” Id., at 1412.

See also id., at 1434-1435 (factual findings). “[I]n the light of this very substantial authority favoring single-sex education,” the District Court concluded that “the VMI Board’s decision to maintain an all-male institution is fully justified even without taking into consideration the other unique features of VMI’s teaching and training.” Id., at 1412. This finding alone, which even this Court cannot dispute, see ante, at 535, should be sufficient to demonstrate the constitutionality of VMI’s all-male composition.

But besides its single-sex constitution, VMI is different from other colleges in another way. It employs a “distinctive educational method,” sometimes referred to as the “adversative, or doubting, model of education.” 766 F. Supp., at 1413, 1421. “Physical rigor, mental stress, absolute equality of treatment, absence of privacy, minute regulation of behavior, and indoctrination in desirable values are the salient attributes of the VMI educational experience.” Id., at 1421. No one contends that this method is appropriate for all individuals; education is not a “one size fits all” business. Just as a State may wish to support junior colleges, vocational institutes, or a law school that emphasizes case 578*578 practice instead of classroom study, so too a State’s decision to maintain within its system one school that provides the adversative method is “substantially related” to its goal of good education. Moreover, it was uncontested that “if the state were to establish a women’s VMI-type [i. e., adversative] program, the program would attract an insufficient number of participants to make the program work,” 44 F. 3d, at 1241; and it was found by the District Court that if Virginia were to include women in VMI, the school “would eventually find it necessary to drop the adversative system altogether,” 766 F. Supp., at 1413. Thus, Virginia’s options were an adversative method that excludes women or no adversative method at all.

There can be no serious dispute that, as the District Court found, single-sex education and a distinctive educational method “represent legitimate contributions to diversity in the Virginia higher education system.” Ibid. As a theoretical matter, Virginia’s educational interest would have been best served (insofar as the two factors we have mentioned are concerned) by six different types of public colleges—an all-men’s, an all-women’s, and a coeducational college run in the “adversative method,” and an all-men’s, an all-women’s, and a coeducational college run in the “traditional method.” But as a practical matter, of course, Virginia’s financial resources, like any State’s, are not limitless, and the Commonwealth must select among the available options. Virginia thus has decided to fund, in addition to some 14 coeducational 4-year colleges, one college that is run as an all-male school on the adversative model: the Virginia Military Institute.

Virginia did not make this determination regarding the make-up of its public college system on the unrealistic assumption that no other colleges exist. Substantial evidence in the District Court demonstrated that the Commonwealth has long proceeded on the principle that “`[h]igher education resources should be viewed as a whole—public and private’ 579*579 “—because such an approach enhances diversity and because “`it is academic and economic waste to permit unwarranted duplication.’ ” Id., at 1420-1421 (quoting 1974 Report of the General Assembly Commission on Higher Education to the General Assembly of Virginia). It is thus significant that, whereas there are “four all-female private [colleges] in Virginia,” there is only “one private all-male college,” which “indicates that the private sector is providing for th[e] [former] form of education to a much greater extent that it provides for all-male education.” 766 F. Supp., at 1420-1421. In these circumstances, Virginia’s election to fund one public all-male institution and one on the adversative model—and to concentrate its resources in a single entity that serves both these interests in diversity—is substantially related to the Commonwealth’s important educational interests.


The Court today has no adequate response to this clear demonstration of the conclusion produced by application of intermediate scrutiny. Rather, it relies on a series of contentions that are irrelevant or erroneous as a matter of law, foreclosed by the record in this litigation, or both.

1. I have already pointed out the Court’s most fundamental error, which is its reasoning that VMI’s all-male composition is unconstitutional because “some women are capable of all of the individual activities required of VMI cadets,” 766 F. Supp., at 1412, and would prefer military training on the adversative model. See supra, at 571-574. This unacknowledged adoption of what amounts to (at least) strict scrutiny is without antecedent in our sex-discrimination cases and by itself discredits the Court’s decision.

2. The Court suggests that Virginia’s claimed purpose in maintaining VMI as an all-male institution—its asserted interest in promoting diversity of educational options—is not “genuin[e],” but is a pretext for discriminating against women. Ante, at 539; see ante, at 535-540. To support this 580*580 charge, the Court would have to impute that base motive to VMI’s Mission Study Committee, which conducted a 3-year study from 1983 to 1986 and recommended to VMI’s Board of Visitors that the school remain all male. The committee, a majority of whose members consisted of non-VMI graduates, “read materials on education and on women in the military,” “made site visits to single-sex and newly coeducational institutions” including West Point and the Naval Academy, and “considered the reasons that other institutions had changed from single-sex to coeducational status”; its work was praised as “thorough” in the accreditation review of VMI conducted by the Southern Association of Colleges and Schools. See 766 F. Supp., at 1413, 1428; see also id., at 1427-1430 (detailed findings of fact concerning the Mission Study Committee). The Court states that “[w]hatever internal purpose the Mission Study Committee served— and however well meaning the framers of the report—we can hardly extract from that effort any commonwealth policy evenhandedly to advance diverse educational options.” Ante, at 539. But whether it is part of the evidence to prove that diversity was the Commonwealth’s objective (its short report said nothing on that particular subject) is quite separate from whether it is part of the evidence to prove that antifeminism was not. The relevance of the Mission Study Committee is that its very creation, its sober 3-year study, and the analysis it produced utterly refute the claim that VMI has elected to maintain its all-male student-body composition for some misogynistic reason.

The Court also supports its analysis of Virginia’s “actual state purposes” in maintaining VMI’s student body as all male by stating that there is no explicit statement in the record “`in which the Commonwealth has expressed itself’ ” concerning those purposes. Ante, at 535, 539 (quoting 976 F. 2d 890, 899 (CA4 1992)); see also ante, at 525. That is wrong on numerous grounds. First and foremost, in its implication that such an explicit statement of “actual purposes” 581*581 is needed. The Court adopts, in effect, the argument of the United States that since the exclusion of women from VMI in 1839 was based on the “assumptions” of the time “that men alone were fit for military and leadership roles,” and since “[b]efore this litigation was initiated, Virginia never sought to supply a valid, contemporary rationale for VMI’s exclusionary policy,” “[t]hat failure itself renders the VMI policy invalid.” Brief for United States in No. 94-2107, at 10. This is an unheard-of doctrine. Each state decision to adopt or maintain a governmental policy need not be accompanied—in anticipation of litigation and on pain of being found to lack a relevant state interest—by a lawyer’s contemporaneous recitation of the State’s purposes. The Constitution is not some giant Administrative Procedure Act, which imposes upon the States the obligation to set forth a “statement of basis and purpose” for their sovereign Acts, see 5 U. S. C. § 553(c). The situation would be different if what the Court assumes to have been the 1839 policy had been enshrined and remained enshrined in legislation—a VMI charter, perhaps, pronouncing that the institution’s purpose is to keep women in their place. But since the 1839 policy was no more explicitly recorded than the Court contends the present one is, the mere fact that today’s Commonwealth continues to fund VMI “is enough to answer [the United States’] contention that the [classification] was the `accidental by-product of a traditional way of thinking about females.’ ” Michael M., 450 U. S., at 471, n. 6 (plurality opinion) (quoting Califano v. Webster, 430 U. S., at 320) (internal quotation marks omitted).

It is, moreover, not true that Virginia’s contemporary reasons for maintaining VMI are not explicitly recorded. It is hard to imagine a more authoritative source on this subject than the 1990 Report of the Virginia Commission on the University of the 21st Century (1990 Report). As the parties stipulated, that report “notes that the hallmarks of Virginia’s educational policy are `diversity and autonomy.’ ” Stipulations 582*582 of Fact 37, reprinted in Lodged Materials from the Record 64 (Lodged Materials). It said: “The formal system of higher education in Virginia includes a great array of institutions: state-supported and independent, two-year and senior, research and highly specialized, traditionally black and single-sex. ” 1990 Report, quoted in relevant part at Lodged Materials 64-65 (emphasis added).[2] The Court’s only response to this is repeated reliance on the Court of Appeals’ assertion that “`the only explicit [statement] that we have found in the record in which the Commonwealth has expressed itself with respect to gender distinctions’ ” (namely, the statement in the 1990 Report that the Commonwealth’s institutions must “deal with faculty, staff, and students without regard to sex”) had nothing to do with the purpose of diversity. Ante, at 525, 539 (quoting 976 F. 2d, at 899). This proves, I suppose, that the Court of Appeals did not find a statement dealing with sex and diversity in the record; but the pertinent question (accepting the need for such a statement) is whether it was there. And the plain fact, which the Court does not deny, is that it was.

583*583 The Court contends that “[a] purpose genuinely to advance an array of educational options . . . is not served” by VMI. Ante, at 539-540. It relies on the fact that all of Virginia’s other public colleges have become coeducational. Ibid.; see also ante, at 521, n. 2. The apparent theory of this argument is that unless Virginia pursues a great deal of diversity, its pursuit of some diversity must be a sham. This fails to take account of the fact that Virginia’s resources cannot support all possible permutations of schools, see supra, at 578, and of the fact that Virginia coordinates its public educational offerings with the offerings of in-state private educational institutions that the Commonwealth provides money for its residents to attend and otherwise assists—which include four women’s colleges.[3]

Finally, the Court unreasonably suggests that there is some pretext in Virginia’s reliance upon decentralized decisionmaking 584*584 to achieve diversity—its granting of substantial autonomy to each institution with regard to student-body composition and other matters, see 766 F. Supp., at 1419. The Court adopts the suggestion of the Court of Appeals that it is not possible for “one institution with autonomy, but with no authority over any other state institution, [to] give effect to a state policy of diversity among institutions.” Ante, at 539 (internal quotation marks omitted). If it were impossible for individual human beings (or groups of human beings) to act autonomously in effective pursuit of a common goal, the game of soccer would not exist. And where the goal is diversity in a free market for services, that tends to be achieved even by autonomous actors who act out of entirely selfish interests and make no effort to cooperate. Each Virginia institution, that is to say, has a natural incentive to make itself distinctive in order to attract a particular segment of student applicants. And of course none of the institutions is entirely autonomous; if and when the legislature decides that a particular school is not well serving the interest of diversity—if it decides, for example, that a men’s school is not much needed—funding will cease.[4]

585*585 3. In addition to disparaging Virginia’s claim that VMI’s single-sex status serves a state interest in diversity, the Court finds fault with Virginia’s failure to offer education based on the adversative training method to women. It dismisses the District Court’s “`findings’ on `gender-based developmental differences’ ” on the ground that “[t]hese `findings’ restate the opinions of Virginia’s expert witnesses, opinions about typically male or typically female `tendencies.’ ” Ante, at 541 (quoting 766 F. Supp., at 1434-1435). How remarkable to criticize the District Court on the ground that its findings rest on the evidence (i. e., the testimony of Virginia’s witnesses)! That is what findings are supposed to do. It is indefensible to tell the Commonwealth that “[t]he burden of justification is demanding and it rests entirely on [you],” ante, at 533, and then to ignore the District Court’s findings because they rest on the evidence put forward by the Commonwealth—particularly when, as the District Court said, “[t]he evidence in the case . . . is virtually uncontradicted, ” 766 F. Supp., at 1415 (emphasis added).

Ultimately, in fact, the Court does not deny the evidence supporting these findings. See ante, at 541-546. It instead makes evident that the parties to this litigation could have saved themselves a great deal of time, trouble, and expense by omitting a trial. The Court simply dispenses with the evidence submitted at trial—it never says that a single finding of the District Court is clearly erroneous—in favor of the Justices’ own view of the world, which the Court proceeds to support with (1) references to observations of someone 586*586 who is not a witness, nor even an educational expert, nor even a judge who reviewed the record or participated in the judgment below, but rather a judge who merely dissented from the Court of Appeals’ decision not to rehear this litigation en banc, see ante, at 542, (2) citations of nonevidentiary materials such as amicus curiae briefs filed in this Court, see ante, at 544-545, nn. 13, 14, and (3) various historical anecdotes designed to demonstrate that Virginia’s support for VMI as currently constituted reminds the Justices of the “bad old days,” see ante, at 542-544.

It is not too much to say that this approach to the litigation has rendered the trial a sham. But treating the evidence as irrelevant is absolutely necessary for the Court to reach its conclusion. Not a single witness contested, for example, Virginia’s “substantial body of `exceedingly persuasive’ evidence . . . that some students, both male and female, benefit from attending a single-sex college” and “[that] [f]or those students, the opportunity to attend a single-sex college is a valuable one, likely to lead to better academic and professional achievement.” 766 F. Supp., at 1411-1412. Even the United States’ expert witness “called himself a `believer in single-sex education,’ ” although it was his “personal, philosophical preference,” not one “born of educational-benefit considerations,” “that single-sex education should be provided only by the private sector.” Id., at 1412.

4. The Court contends that Virginia, and the District Court, erred, and “misperceived our precedent,” by “train[ing] their argument on `means’ rather than `end,’ ” ante, at 545. The Court focuses on “VMI’s mission,” which is to produce individuals “imbued with love of learning, confident in the functions and attitudes of leadership, possessing a high sense of public service, advocates of the American democracy and free enterprise system, and ready . . . to defend their country in time of national peril.” 766 F. Supp., at 1425 (quoting Mission Study Committee of the VMI Board of 587*587 Visitors, Report, May 16, 1986). “Surely,” the Court says, “that goal is great enough to accommodate women.” Ante, at 545.

This is lawmaking by indirection. What the Court describes as “VMI’s mission” is no less the mission of all Virginia colleges. Which of them would the Old Dominion continue to fund if they did not aim to create individuals “imbued with love of learning, etc.,” right down to being ready “to defend their country in time of national peril”? It can be summed up as “learning, leadership, and patriotism.” To be sure, those general educational values are described in a particularly martial fashion in VMI’s mission statement, in accordance with the military, adversative, and all-male character of the institution. But imparting those values in that fashioni. e., in a military, adversative, all-male environment—is the distinctive mission of VMI. And as I have discussed (and both courts below found), that mission is not “great enough to accommodate women.”

The Court’s analysis at least has the benefit of producing foreseeable results. Applied generally, it means that whenever a State’s ultimate objective is “great enough to accommodate women” (as it always will be), then the State will be held to have violated the Equal Protection Clause if it restricts to men even one means by which it pursues that objective—no matter how few women are interested in pursuing the objective by that means, no matter how much the single-sex program will have to be changed if both sexes are admitted, and no matter how beneficial that program has theretofore been to its participants.

5. The Court argues that VMI would not have to change very much if it were to admit women. See, e. g., ante, at 540-542. The principal response to that argument is that it is irrelevant: If VMI’s single-sex status is substantially related to the government’s important educational objectives, as I have demonstrated above and as the Court refuses to discuss, 588*588 that concludes the inquiry. There should be no debate in the federal judiciary over “how much” VMI would be required to change if it admitted women and whether that would constitute “too much” change.

But if such a debate were relevant, the Court would certainly be on the losing side. The District Court found as follows: “[T]he evidence establishes that key elements of the adversative VMI educational system, with its focus on barracks life, would be fundamentally altered, and the distinctive ends of the system would be thwarted, if VMI were forced to admit females and to make changes necessary to accommodate their needs and interests.” 766 F. Supp., at 1411. Changes that the District Court’s detailed analysis found would be required include new allowances for personal privacy in the barracks, such as locked doors and coverings on windows, which would detract from VMI’s approach of regulating minute details of student behavior, “contradict the principle that everyone is constantly subject to scrutiny by everyone else,” and impair VMI’s “total egalitarian approach” under which every student must be “treated alike”; changes in the physical training program, which would reduce “[t]he intensity and aggressiveness of the current program”; and various modifications in other respects of the adversative training program that permeates student life. See id., at 1412-1413, 1435-1443. As the Court of Appeals summarized it, “the record supports the district court’s findings that at least these three aspects of VMI’s program— physical training, the absence of privacy, and the adversative approach—would be materially affected by coeducation, leading to a substantial change in the egalitarian ethos that is a critical aspect of VMI’s training.” 976 F. 2d, at 896-897.

In the face of these findings by two courts below, amply supported by the evidence, and resulting in the conclusion that VMI would be fundamentally altered if it admitted women, this Court simply pronounces that “[t]he notion that 589*589 admission of women would downgrade VMI’s stature, destroy the adversative system and, with it, even the school, is a judgment hardly proved.” Ante, at 542 (footnote omitted). The point about “downgrad[ing] VMI’s stature” is a straw man; no one has made any such claim. The point about “destroy[ing] the adversative system” is simply false; the District Court not only stated that “[e]vidence supports this theory,” but specifically concluded that while “[w]ithout a doubt” VMI could assimilate women, “it is equally without a doubt that VMI’s present methods of training and education would have to be changed” by a “move away from its adversative new cadet system.” 766 F. Supp., at 1413, and n. 8, 1440. And the point about “destroy[ing] the school,” depending upon what that ambiguous phrase is intended to mean, is either false or else sets a standard much higher than VMI had to meet. It sufficed to establish, as the District Court stated, that VMI would be “significantly different” upon the admission of women, 766 F. Supp., at 1412, and “would eventually find it necessary to drop the adversative system altogether,” id., at 1413.[5]

590*590 6. Finally, the absence of a precise “all-women’s analogue” to VMI is irrelevant. In Mississippi Univ. for Women v. Hogan, 458 U. S. 718 (1982), we attached no constitutional significance to the absence of an all-male nursing school. As Virginia notes, if a program restricted to one sex is necessarily unconstitutional unless there is a parallel program restricted to the other sex, “the opinion in Hogan could have ended with its first footnote, which observed that `Mississippi maintains no other single-sex public university or college.’ ” Brief for Cross-Petitioners in No. 94-2107, p. 38 (quoting Mississippi Univ. for Women v. Hogan, supra, at 720, n. 1).

Although there is no precise female-only analogue to VMI, Virginia has created during this litigation the Virginia Women’s Institute for Leadership (VWIL), a state-funded all-women’s program run by Mary Baldwin College. I have thus far said nothing about VWIL because it is, under our established test, irrelevant, so long as VMI `s all-male character is “substantially related” to an important state goal. But VWIL now exists, and the Court’s treatment of it shows how far reaching today’s decision is.

VWIL was carefully designed by professional educators who have long experience in educating young women. The program rejects the proposition that there is a “difference in the respective spheres and destinies of man and woman,” Bradwell v. State, 16 Wall. 130, 141 (1873), and is designed to “provide an all-female program that will achieve substantially similar outcomes [to VMI’s] in an all-female environment,” 852 F. Supp. 471, 481 (WD Va. 1994). After holding a trial where voluminous evidence was submitted and making detailed findings of fact, the District Court concluded that “there is a legitimate pedagogical basis for the different means employed [by VMI and VWIL] to achieve the substantially 591*591 similar ends.” Ibid. The Court of Appeals undertook a detailed review of the record and affirmed. 44 F. 3d 1229 (CA4 1995).[6] But it is Mary Baldwin College, which runs VWIL, that has made the point most succinctly:

“It would have been possible to develop the VWIL program to more closely resemble VMI, with adversative techniques associated with the rat line and barracks-like living quarters. Simply replicating an existing program would have required far less thought, research, and educational expertise. But such a facile approach would have produced a paper program with no real prospect of successful implementation.” Brief for Mary Baldwin College as Amicus Curiae 5.

It is worth noting that none of the United States’ own experts in the remedial phase of this litigation was willing to testify that VMI’s adversative method was an appropriate methodology for educating women. This Court, however, does not care. Even though VWIL was carefully designed by professional educators who have tremendous experience in the area, and survived the test of adversarial litigation, the Court simply declares, with no basis in the evidence, that 592*592 these professionals acted on “`overbroad’ generalizations,” ante, at 542, 550.


A few words are appropriate in response to the concurrence, which finds VMI unconstitutional on a basis that is more moderate than the Court’s but only at the expense of being even more implausible. The concurrence offers three reasons: First, that there is “scant evidence in the record,” ante, at 562, that diversity of educational offering was the real reason for Virginia’s maintaining VMI. “Scant” has the advantage of being an imprecise term. I have cited the clearest statements of diversity as a goal for higher education in the 1990 Report, the 1989 Virginia Plan for Higher Education, the Budget Initiatives prepared in 1989 by the State Council of Higher Education for Virginia, the 1974 Report of the General Assembly Commission on Higher Education to the General Assembly of Virginia, and the 1969 Report of the Virginia Commission on Constitutional Revision. See supra, at 579, 581-582, and n. 2, 583, n. 3. There is no evidence to the contrary, once one rejects (as the concurrence rightly does) the relevance of VMI’s founding in days when attitudes toward the education of women were different. Is this conceivably not enough to foreclose rejecting as clearly erroneous the District Court’s determination regarding “the Commonwealth’s objective of educational diversity”? 766 F. Supp., at 1413. Especially since it is absurd on its face even to demand “evidence” to prove that the Commonwealth’s reason for maintaining a men’s military academy is that a men’s military academy provides a distinctive type of educational experience (i. e., fosters diversity). What other purpose would the Commonwealth have? One may argue, as the Court does, that this type of diversity is designed only to indulge hostility toward women—but that is a separate point, explicitly rejected by the concurrence, and amply refuted by the evidence I have mentioned in discussing 593*593 the Court’s opinion.[7] What is now under discussion—the concurrence’s making central to the disposition of this litigation the supposedly “scant” evidence that Virginia maintained VMI in order to offer a diverse educational experience—is rather like making crucial to the lawfulness of the United States Army record “evidence” that its purpose is to do battle. A legal culture that has forgotten the concept of res ipsa loquitur deserves the fate that it today decrees for VMI.

Second, the concurrence dismisses out of hand what it calls Virginia’s “second justification for the single-sex admissions policy: maintenance of the adversative method.” Ante, at 564. The concurrence reasons that “this justification does not serve an important governmental objective” because, whatever the record may show about the pedagogical benefits of single-sex education, “there is no similar evidence in the record that an adversative method is pedagogically beneficial or is any more likely to produce character traits than other methodologies.” Ibid. That is simply wrong. See, e. g., 766 F. Supp., at 1426 (factual findings concerning character traits produced by VMI’s adversative methodology); id. , at 1434 (factual findings concerning benefits for many college-age men of an adversative approach in general). In reality, the pedagogical benefits of VMI’s adversative approach were not only proved, but were a given in this litigation. The reason the woman applicant who prompted this suit wanted to enter VMI was assuredly not that she wanted to go to an all-male school; it would cease being all-male as 594*594 soon as she entered. She wanted the distinctive adversative education that VMI provided, and the battle was joined (in the main) over whether VMI had a basis for excluding women from that approach. The Court’s opinion recognizes this, and devotes much of its opinion to demonstrating that “`some women . . . do well under [the] adversative model’ ” and that “[i]t is on behalf of these women that the United States has instituted this suit.” Ante, at 550 (quoting 766 F. Supp., at 1434). Of course, in the last analysis it does not matter whether there are any benefits to the adversative method. The concurrence does not contest that there are benefits to single-sex education, and that alone suffices to make Virginia’s case, since admission of a woman will even more surely put an end to VMI’s single-sex education than it will to VMI’s adversative methodology.

A third reason the concurrence offers in support of the judgment is that the Commonwealth and VMI were not quick enough to react to the “further developments” in this Court’s evolving jurisprudence. Ante, at 561. Specifically, the concurrence believes it should have been clear after Hogan that “[t]he difficulty with [Virginia’s] position is that the diversity benefited only one sex; there was single-sex public education available for men at VMI, but no corresponding single-sex public education available for women.” Ante, at 562. If only, the concurrence asserts, Virginia had “made a genuine effort to devote comparable public resources to a facility for women, and followed through on such a plan, it might well have avoided an equal protection violation.” Ante, at 563. That is to say, the concurrence believes that after our decision in Hogan (which held a program of the Mississippi University for Women to be unconstitutional—without any reliance on the fact that there was no corresponding Mississippi all-men’s program), the Commonwealth should have known that what this Court expected of it was . . . yes!, the creation of a state all-women’s program. Any lawyer who gave that advice to the Commonwealth 595*595 ought to have been either disbarred or committed. (The proof of that pudding is today’s 6-Justice majority opinion.) And any Virginia politician who proposed such a step when there were already four 4-year women’s colleges in Virginia (assisted by state support that may well exceed, in the aggregate, what VMI costs, see n. 3, supra ) ought to have been recalled.

In any event, “diversity in the form of single-sex, as well as coeducational, institutions of higher learning” is “available to women as well as to men” in Virginia. Ante, at 564. The concurrence is able to assert the contrary only by disregarding the four all-women’s private colleges in Virginia (generously assisted by public funds) and the Commonwealth’s longstanding policy of coordinating public with private educational offerings, see supra, at 579, 581-582, and n. 2, 583-584, and n. 3. According to the concurrence, the

2, Virginia’s reason assistance to its four all-women’s private colleges does not count is that “[t]he private women’s colleges are treated by the State exactly as all other private schools are treated.” Ante, at 564. But if Virginia cannot get credit for assisting women’s education if it only treats women’s private schools as it does all other private schools, then why should it get blame for assisting men’s education if it only treats VMI as it does all other public schools? This is a great puzzlement.


As is frequently true, the Court’s decision today will have consequences that extend far beyond the parties to the litigation. What I take to be the Court’s unease with these consequences, and its resulting unwillingness to acknowledge them, cannot alter the reality.


Under the constitutional principles announced and applied today, single-sex public education is unconstitutional. By going through the motions of applying a balancing test—asking 596*596 whether the State has adduced an “exceedingly persuasive justification” for its sex-based classification—the Court creates the illusion that government officials in some future case will have a clear shot at justifying some sort of singlesex public education. Indeed, the Court seeks to create even a greater illusion than that: It purports to have said nothing of relevance to other public schools at all. “We address specifically and only an educational opportunity recognized . . . as `unique.’ ” Ante, at 534, n. 7.

The Supreme Court of the United States does not sit to announce “unique” dispositions. Its principal function is to establish precedent —that is, to set forth principles of law that every court in America must follow. As we said only this Term, we expect both ourselves and lower courts to adhere to the “rationale upon which the Court based the results of its earlier decisions.” Seminole Tribe of Fla. v. Florida, 517 U. S. 44, 66-67 (1996) (emphasis added). That is the principal reason we publish our opinions.

And the rationale of today’s decision is sweeping: for sexbased classifications, a redefinition of intermediate scrutiny that makes it indistinguishable from strict scrutiny. See supra, at 571-574. Indeed, the Court indicates that if any program restricted to one sex is “uniqu[e],” it must be opened to members of the opposite sex “who have the will and capacity” to participate in it. Ante, at 542. I suggest that the single-sex program that will not be capable of being characterized as “unique” is not only unique but nonexistent.[8]

In any event, regardless of whether the Court’s rationale leaves some small amount of room for lawyers to argue, it ensures that single-sex public education is functionally dead. 597*597 The costs of litigating the constitutionality of a single-sex education program, and the risks of ultimately losing that litigation, are simply too high to be embraced by public officials. Any person with standing to challenge any sex-based classification can haul the State into federal court and compel it to establish by evidence (presumably in the form of expert testimony) that there is an “exceedingly persuasive justification” for the classification. Should the courts happen to interpret that vacuous phrase as establishing a standard that is not utterly impossible of achievement, there is considerable risk that whether the standard has been met will not be determined on the basis of the record evidence—indeed, that will necessarily be the approach of any court that seeks to walk the path the Court has trod today. No state official in his right mind will buy such a high-cost, high-risk lawsuit by commencing a single-sex program. The enemies of singlesex education have won; by persuading only seven Justices (five would have been enough) that their view of the world is enshrined in the Constitution, they have effectively imposed that view on all 50 States.

This is especially regrettable because, as the District Court here determined, educational experts in recent years have increasingly come to “suppor[t] [the] view that substantial educational benefits flow from a single-gender environment, be it male or female, that cannot be replicated in a coeducational setting. ” 766 F. Supp., at 1415 (emphasis added). “The evidence in th[is] case,” for example, “is virtually uncontradicted” to that effect. Ibid. Until quite recently, some public officials have attempted to institute new single-sex programs, at least as experiments. In 1991, for example, the Detroit Board of Education announced a program to establish three boys-only schools for inner-city youth; it was met with a lawsuit, a preliminary injunction was swiftly entered by a District Court that purported to rely on Hogan, see Garrett v. Board of Ed. of School Dist. of Detroit, 775 F. Supp. 1004, 1006 (ED Mich. 1991), and the 598*598 Detroit Board of Education voted to abandon the litigation and thus abandon the plan, see Detroit Plan to Aid Blacks with All-Boy Schools Abandoned, Los Angeles Times, Nov. 8, 1991, p. A4, col. 1. Today’s opinion assures that no such experiment will be tried again.


There are few extant single-sex public educational programs. The potential of today’s decision for widespread disruption of existing institutions lies in its application to private single-sex education. Government support is immensely important to private educational institutions. Mary Baldwin College—which designed and runs VWIL— notes that private institutions of higher education in the 1990-1991 school year derived approximately 19 percent of their budgets from federal, state, and local government funds, not including financial aid to students. See Brief for Mary Baldwin College as Amicus Curiae 22, n. 13 (citing U. S. Dept. of Education, National Center for Education Statistics, Digest of Education Statistics, p. 38 and Note (1993)). Charitable status under the tax laws is also highly significant for private educational institutions, and it is certainly not beyond the Court that rendered today’s decision to hold that a donation to a single-sex college should be deemed contrary to public policy and therefore not deductible if the college discriminates on the basis of sex. See Note, The Independent Sector and the Tax Laws: Defining Charity in an Ideal Democracy, 64 S. Cal. L. Rev. 461, 476 (1991). See also Bob Jones Univ. v. United States, 461 U. S. 574 (1983).

The Court adverts to private single-sex education only briefly, and only to make the assertion (mentioned above) that “[w]e address specifically and only an educational opportunity recognized by the District Court and the Court of Appeals as `unique.’ ” Ante, at 534, n. 7. As I have already remarked, see supra, at 596, that assurance assures nothing, unless it is to be taken as a promise that in the future 599*599 the Court will disclaim the reasoning it has used today to destroy VMI. The Government, in its briefs to this Court, at least purports to address the consequences of its attack on VMI for public support of private single-sex education. It contends that private colleges that are the direct or indirect beneficiaries of government funding are not thereby necessarily converted into state actors to which the Equal Protection Clause is then applicable. See Brief for United States in No. 94-2107, at 35-37 (discussing Rendell-Baker v. Kohn, 457 U. S. 830 (1982), and Blum v. Yaretsky, 457 U. S. 991 (1982)). That is true. It is also virtually meaningless.

The issue will be not whether government assistance turns private colleges into state actors, but whether the government itself would be violating the Constitution by providing state support to single-sex colleges. For example, in Norwood v. Harrison, 413 U. S. 455 (1973), we saw no room to distinguish between state operation of racially segregated schools and state support of privately run segregated schools. “Racial discrimination in state-operated schools is barred by the Constitution and `[i]t is also axiomatic that a state may not induce, encourage or promote private persons to accomplish what it is constitutionally forbidden to accomplish.’ ” Id., at 465 (quoting Lee v. Macon County Bd. of Ed., 267 F. Supp. 458, 475-476 (MD Ala. 1967)); see also Cooper v. Aaron, 358 U. S. 1, 19 (1958) (“State support of segregated schools through any arrangement, management, funds, or property cannot be squared with the [Fourteenth] Amendment’s command that no State shall deny to any person within its jurisdiction the equal protection of the laws”); Grove City College v. Bell, 465 U. S. 555, 565 (1984) (case arising under Title IX of the Education Amendments of 1972 and stating that “[t]he economic effect of direct and indirect assistance often is indistinguishable”). When the Government was pressed at oral argument concerning the implications of these cases for private single-sex education if government-provided single-sex education is unconstitutional, 600*600 it stated that the implications will not be so disastrous, since States can provide funding to racially segregated private schools, “depend[ing] on the circumstances,” Tr. of Oral Arg. 56. I cannot imagine what those “circumstances” might be, and it would be as foolish for privateschool administrators to think that that assurance from the Justice Department will outlive the day it was made, as it was for VMI to think that the Justice Department’s “unequivoca[l]” support for an intermediate-scrutiny standard in this litigation would survive the Government’s loss in the courts below.

The only hope for state-assisted single-sex private schools is that the Court will not apply in the future the principles of law it has applied today. That is a substantial hope, I am happy and ashamed to say. After all, did not the Court today abandon the principles of law it has applied in our earlier sex-classification cases? And does not the Court positively invite private colleges to rely upon our ad-hocery by assuring them this litigation is “unique”? I would not advise the foundation of any new single-sex college (especially an all-male one) with the expectation of being allowed to receive any government support; but it is too soon to abandon in despair those single-sex colleges already in existence. It will certainly be possible for this Court to write a future opinion that ignores the broad principles of law set forth today, and that characterizes as utterly dispositive the opinion’s perceptions that VMI was a uniquely prestigious allmale institution, conceived in chauvinism, etc., etc. I will not join that opinion.

* * *

Justice Brandeis said it is “one of the happy incidents of the federal system that a single courageous State may, if its citizens choose, serve as a laboratory; and try novel social and economic experiments without risk to the rest of the country.” New State Ice Co. v. Liebmann, 285 U. S. 262, 311 601*601 (1932) (dissenting opinion). But it is one of the unhappy incidents of the federal system that a self-righteous Supreme Court, acting on its Members’ personal view of what would make a “`more perfect Union,’ ” ante, at 558 (a criterion only slightly more restrictive than a “more perfect world”), can impose its own favored social and economic dispositions nationwide. As today’s disposition, and others this single Term, show, this places it beyond the power of a “single courageous State,” not only to introduce novel dispositions that the Court frowns upon, but to reintroduce, or indeed even adhere to, disfavored dispositions that are centuries old. See, e. g., BMW of North America, Inc. v. Gore, 517 U. S. 559 (1996); Romer v. Evans, 517 U. S. 620 (1996). The sphere of self-government reserved to the people of the Republic is progressively narrowed.

In the course of this dissent, I have referred approvingly to the opinion of my former colleague, Justice Powell, in Mississippi Univ. for Women v. Hogan, 458 U. S. 718 (1982). Many of the points made in his dissent apply with equal force here—in particular, the criticism of judicial opinions that purport to be “narro[w]” but whose “logic” is “sweepin[g].” Id., at 745-746, n. 18. But there is one statement with which I cannot agree. Justice Powell observed that the Court’s decision in Hogan, which struck down a single-sex program offered by the Mississippi University for Women, had thereby “[l]eft without honor . . . an element of diversity that has characterized much of American education and enriched much of American life.” Id., at 735. Today’s decision does not leave VMI without honor; no court opinion can do that.

In an odd sort of way, it is precisely VMI’s attachment to such old-fashioned concepts as manly “honor” that has made it, and the system it represents, the target of those who today succeed in abolishing public single-sex education. The record contains a booklet that all first-year VMI students 602*602 (the so-called “rats”) were required to keep in their possession at all times. Near the end there appears the following period piece, entitled “The Code of a Gentleman”:

“Without a strict observance of the fundamental Code of Honor, no man, no matter how `polished,’ can be considered a gentleman. The honor of a gentleman demands the inviolability of his word, and the incorruptibility of his principles. He is the descendant of the knight, the crusader; he is the defender of the defenseless and the champion of justice . . . or he is not a Gentleman.

“A Gentleman . . .

“Does not discuss his family affairs in public or with acquaintances.

“Does not speak more than casually about his girl friend.

“Does not go to a lady’s house if he is affected by alcohol. He is temperate in the use of alcohol.

“Does not lose his temper; nor exhibit anger, fear, hate, embarrassment, ardor or hilarity in public.

“Does not hail a lady from a club window.

“A gentleman never discusses the merits or demerits of a lady.

“Does not mention names exactly as he avoids the mention of what things cost.

“Does not borrow money from a friend, except in dire need. Money borrowed is a debt of honor, and must be repaid as promptly as possible. Debts incurred by a deceased parent, brother, sister or grown child are assumed by honorable men as a debt of honor.

“Does not display his wealth, money or possessions.

“Does not put his manners on and off, whether in the club or in a ballroom. He treats people with courtesy, no matter what their social position may be.

603*603 “Does not slap strangers on the back nor so much as lay a finger on a lady.

“Does not `lick the boots of those above’ nor `kick the face of those below him on the social ladder.’

“Does not take advantage of another’s helplessness or ignorance and assumes that no gentleman will take advantage of him.

“A Gentleman respects the reserves of others, but demands that others respect those which are his.

“A Gentleman can become what he wills to be. . . .”

I do not know whether the men of VMI lived by this code; perhaps not. But it is powerfully impressive that a public institution of higher education still in existence sought to have them do so. I do not think any of us, women included, will be better off for its destruction.

[*] Together with No. 94-2107, Virginia et al. v. United States, also on certiorari to the same court.

[†] Briefs of amici curiae urging reversal in No. 94-1941 were filed for the State of Maryland et al. by J. Joseph Curran, Jr., Attorney General of Maryland, and Andrew H. Baida, Assistant Attorney General, and by the Attorneys General for their respective jurisdictions as follows: Margery S. Bronster of Hawaii, Scott Harshbarger of Massachusetts, Frankie Sue Del Papa of Nevada, C. Sebastian Aloot of the Northern Mariana Islands, and Theodore R. Kulongoski of Oregon; for the Employment Law Center et al. by Patricia A. Shiu and Judith Kurtz; and for the National Women’s Law Center et al. by Robert N. Weiner, Marcia D. Greenberger, Sara L. Mandelbaum, Janet Gallagher, Mary Wyckoff, Steven R. Shapiro, and Susan Deller Ross.

Briefs of amici curiae urging affirmance in No. 94-1941 were filed for the State of South Carolina et al. by Charles Molony Condon, Attorney General, Treva Ashworth, Deputy Attorney General, Kenneth P. Woodington, Senior Assistant Attorney General, Reginald I. Lloyd, Assistant Attorney General, and M. Dawes Cooke, Jr.; and for Kenneth E. Clark et al. by James C. Roberts and George A. Somerville.

Briefs of amici curiae were filed in both cases for the State of Wyoming et al. by William U. Hill, Attorney General of Wyoming, Thomas W. Corbett, Jr., Attorney General of Pennsylvania, and Bradley B. Cavedo; for Bennett College et al. by Wendy S. White; for the Center for Military Readiness et al. by Mellissa Wells-Petry and Jordan W. Lorence; for the Employment Law Center et al. by Patricia A. Shiu and Judith Kurtz; for the Independent Women’s Forum et al. by Anita K. Blair and C. Douglas Welty; for Mary Baldwin College by Craig T. Merritt and Richard K. Willard; for the South Carolina Institute of Leadership for Women by Julianne Farnsworth; for Wells College et al. by David M. Lascell; for Women’s Schools Together, Inc., et al. by John C. Danforth and Thomas C. Walsh; and for Nancy Mellette by Valorie K. Vojdik, Henry Weisburg, Suzanne E. Coe, and Robert R. Black.

Briefs of amici curiae were filed in No. 94-1941 for the American Association of University Professors et al. by Joan E. Bertin and Ann H. Franke; and for Rhonda Cornum et al. by Allan L. Gropper.

Daniel F. Kolb, Herbert J. Hansell, Paul C. Saunders, Norman Redlich, Barbara R. Arnwine, Thomas J. Henderson, and Richard T. Seymour filed a brief for the Lawyers’ Committee for Civil Rights Under Law as amicus curiae in No. 94-2107.

[1] During the Civil War, school teaching became a field dominated by women. See A. Scott, The Southern Lady: From Pedestal to Politics, 1830-1930, p. 82 (1970).

[2] Historically, most of Virginia’s public colleges and universities were single sex; by the mid-1970’s, however, all except VMI had become coeducational. 766 F. Supp. 1407, 1418-1419 (WD Va. 1991). For example, Virginia’s legislature incorporated Farmville Female Seminary Association in 1839, the year VMI opened. 1839 Va. Acts, ch. 167. Originally providing instruction in “English, Latin, Greek, French, and piano” in a “home atmosphere,” R. Sprague, Longwood College: A History 7-8, 15 (1989) (Longwood College), Farmville Female Seminary became a public institution in 1884 with a mission to train “white female teachers for public schools,” 1884 Va. Acts, ch. 311. The school became Longwood College in 1949, Longwood College 136, and introduced coeducation in 1976, id. , at 133.

[3] The District Court allowed the VMI Foundation and the VMI Alumni Association to intervene as defendants. 766 F. Supp., at 1408.

[4] Six judges voted to rehear the case en banc, four voted against rehearing, and three were recused. The Fourth Circuit’s local Rule permits rehearing en banc only on the vote of a majority of the Circuit’s judges in regular active service (currently 13) without regard to recusals. See 52 F. 3d, at 91, and n. 1.

[5] As Thomas Jefferson stated the view prevailing when the Constitution was new:

“Were our State a pure democracy . . . there would yet be excluded from their deliberations . . . [w]omen, who, to prevent depravation of morals and ambiguity of issue, could not mix promiscuously in the public meetings of men.” Letter from Thomas Jefferson to Samuel Kercheval (Sept. 5, 1816), in 10 Writings of Thomas Jefferson 45-46, n. 1 (P. Ford ed. 1899).

[6] The Court has thus far reserved most stringent judicial scrutiny for classifications based on race or national origin, but last Term observed that strict scrutiny of such classifications is not inevitably “fatal in fact.” Adarand Constructors, Inc. v. Peña, 515 U. S. 200, 237 (1995) (internal quotation marks omitted).

[7] Several amici have urged that diversity in educational opportunities is an altogether appropriate governmental pursuit and that single-sex schools can contribute importantly to such diversity. Indeed, it is the mission of some single-sex schools “to dissipate, rather than perpetuate, traditional gender classifications.” See Brief for Twenty-six Private Women’s Colleges as Amici Curiae 5. We do not question the Commonwealth’s prerogative evenhandedly to support diverse educational opportunities. We address specifically and only an educational opportunity recognized by the District Court and the Court of Appeals as “unique,” see 766 F. Supp., at 1413, 1432; 976 F. 2d, at 892, an opportunity available only at Virginia’s premier military institute, the Commonwealth’s sole single-sex public university or college. Cf. Mississippi Univ. for Women v. Hogan, 458 U. S. 718, 720, n. 1 (1982) (“Mississippi maintains no other single-sex public university or college. Thus, we are not faced with the question of whether States can provide `separate but equal’ undergraduate institutions for males and females.”).

[8] On this point, the dissent sees fire where there is no flame. See post, at 596-598, 598-600. “Both men and women can benefit from a single-sex education,” the District Court recognized, although “the beneficial effects” of such education, the court added, apparently “are stronger among women than among men.” 766 F. Supp., at 1414. The United States does not challenge that recognition. Cf. C. Jencks & D. Riesman, The Academic Revolution 297-298 (1968):

“The pluralistic argument for preserving all-male colleges is uncomfortably similar to the pluralistic argument for preserving all-white colleges . . . . The all-male college would be relatively easy to defend if it emerged from a world in which women were established as fully equal to men. But it does not. It is therefore likely to be a witting or unwitting device for preserving tacit assumptions of male superiority—assumptions for which women must eventually pay.”

[9] Dr. Edward H. Clarke of Harvard Medical School, whose influential book, Sex in Education, went through 17 editions, was perhaps the most well-known speaker from the medical community opposing higher education for women. He maintained that the physiological effects of hard study and academic competition with boys would interfere with the development of girls’ reproductive organs. See E. Clarke, Sex in Education 38-39, 62-63 (1873); id. , at 127 (“identical education of the two sexes is a crime before God and humanity, that physiology protests against, and that experience weeps over”); see also H. Maudsley, Sex in Mind and in Education 17 (1874) (“It is not that girls have not ambition, nor that they fail generally to run the intellectual race [in coeducational settings], but it is asserted that they do it at a cost to their strength and health which entails life-long suffering, and even incapacitates them for the adequate performance of the natural functions of their sex.”); C. Meigs, Females and Their Diseases 350 (1848) (after five or six weeks of “mental and educational discipline,” a healthy woman would “lose . . . the habit of menstruation” and suffer numerous ills as a result of depriving her body for the sake of her mind).

[10] Virginia’s Superintendent of Public Instruction dismissed the coeducational idea as “`repugnant to the prejudices of the people’ ” and proposed a female college similar in quality to Girton, Smith, or Vassar.2 History of Women’s Education 254 (quoting Dept. of Interior, 1 Report of Commissioner of Education, H. R. Doc. No. 5, 58th Cong., 2d Sess., 438 (1904)).

[11] See post, at 566, 598-599, 603. Forecasts of the same kind were made regarding admission of women to the federal military academies. See, e. g., Hearings on H. R. 9832 et al. before Subcommittee No. 2 of the House Committee on Armed Services, 93d Cong., 2d Sess., 137 (1975) (statement of Lt. Gen. A. P. Clark, Superintendent of U. S. Air Force Academy) (“It is my considered judgment that the introduction of female cadets will inevitably erode this vital atmosphere.”); id., at 165 (statement of Hon. H. H. Callaway, Secretary of the Army) (“Admitting women to West Point would irrevocably change the Academy. . . . The Spartan atmosphere— which is so important to producing the final product—would surely be diluted, and would in all probability disappear.”).

[12] See 766 F. Supp., at 1413 (describing testimony of expert witness David Riesman: “[I]f VMI were to admit women, it would eventually find it necessary to drop the adversative system altogether, and adopt a system that provides more nurturing and support for the students.”). Such judgments have attended, and impeded, women’s progress toward full citizenship stature throughout our Nation’s history. Speaking in 1879 in support of higher education for females, for example, Virginia State Senator C. T. Smith of Nelson recounted that legislation proposed to protect the property rights of women had encountered resistance. 10 Educ. J. Va. 213 (1879). A Senator opposing the measures objected that “there [was] no formal call for the [legislation],” and “depicted in burning eloquence the terrible consequences such laws would produce.” Ibid. The legislation passed, and a year or so later, its sponsor, C. T. Smith, reported that “not one of [the forecast “terrible consequences”] has or ever will happen, even unto the sounding of Gabriel’s trumpet.” Ibid. See also supra, at 537-538.

[13] Women cadets have graduated at the top of their class at every federal military academy. See Brief for Lieutenant Colonel Rhonda Cornum et al. as Amici Curiae 11, n. 25; cf. Defense Advisory Committee on Women in the Services, Report on the Integration and Performance of Women at West Point 64 (1992).

[14] Brief for Lieutenant Colonel Rhonda Cornum, supra, at 5-9 (reporting the vital contributions and courageous performance of women in the military); see Mintz, President Nominates 1st Woman to Rank of Three-Star General, Washington Post, Mar. 27, 1996, p. A19, col. 1 (announcing President’s nomination of Marine Corps Major General Carol Mutter to rank of Lieutenant General; Mutter will head corps manpower and planning); Tousignant, A New Era for the Old Guard, Washington Post, Mar. 23, 1996, p. C1, col. 2 (reporting admission of Sergeant Heather Johnsen to eliteInfantry unit that keeps round-the-clock vigil at Tomb of the Unknowns in Arlington National Cemetery).

[15] Inclusion of women in settings where, traditionally,they were not wanted inevitably entails a period of adjustment. As one West Point cadet squad leader recounted: “[T]he classes of `78 and `79 see the women as women, but the classes of `80and `81 see them as classmates.” U. S. Military Academy, A. Vitters, Report of Admission of Women (Project Athena II) 84 (1978) (internal quotation marks omitted).

[16] VMI has successfully managed another notable change. The school admitted its first African-American cadets in 1968. See The VMI Story 347-349 (students no longer sing “Dixie,” salute the Confederate flag or the tomb of General Robert E. Lee at ceremonies and sports events). As the District Court noted, VMI established a program on “retention of black cadets” designed to offer academic and social-cultural support to “minority members of a dominantly white and tradition-oriented student body.” 766 F. Supp., at 1436-1437. The school maintains a “special recruitment program for blacks” which, the District Court found, “has had little,if any, effect on VMI’s method of accomplishing its mission.” Id., at 1437.

[17] As earlier observed, see supra, at 529, Judge Phillips, in dissent, measured Virginia’s plan against a paradigm arrangement, one that “could survive equal protection scrutiny”: single-sex schools with “substantially comparable curricular and extra-curricular programs, funding, physical plant, administration and support services, . . . faculty[,] and library resources.” 44 F. 3d 1229, 1250 (CA4 1995). Cf. Bray v. Lee, 337 F. Supp. 934 (Mass. 1972) (holding inconsistent with the Equal Protection Clause admission of males to Boston’s Boys Latin School with a test score of 120 or higher (up to a top score of 200) while requiring a score, on the same test, of at least 133 for admission of females to Girls Latin School, but not ordering coeducation). Measuring VMI/VWIL against the paradigm, Judge Phillips said, “reveals how far short the [Virginia] plan falls from providing substantially equal tangible and intangible educational benefits to men and women.” 44 F. 3d, at 1250.

[18] Both programs include an honor system. Students at VMI are expelled forthwith for honor code violations, see 766 F. Supp., at 1423; the system for VWIL students, see 852 F. Supp., at 496-497, is less severe, see Tr. 414-415 (testimony of Mary Baldwin College President Cynthia Tyson).

[19] Admitting women to VMI would undoubtedly require alterations necessary to afford members of each sex privacy from the other sex in living arrangements, and to adjust aspects of the physical training programs. See Brief for Petitioner 27-29; cf.note following 10 U. S. C. § 4342 (academic and other standards for women admitted to the Military, Naval, and Air Force Academies “shall be the same as those required for male individuals, except for those minimum essential adjustments in such standards required because of physiological differences between male and female individuals”). Experience shows such adjustments are manageable. See U. S. Military Academy, A. Vitters, N. Kinzer, & J. Adams, Report of Admission of Women (Project Athena I-IV) (1977-1980) (4-year longitudinal study of the admission of women to West Point); Defense Advisory Committee on Women in the Services, Report on the Integration and Performance of Women at West Point 17-18 (1992).

[20] Virginia’s prime concern, it appears, is that “plac[ing] men and women into the adversative relationship inherent in the VMI program . . . would destroy, at least for that period of the adversative training, any sense of decency that still permeates the relationship between the sexes.” 44 F. 3d, at 1239; see supra, at 540-546. It is an ancient and familiar fear. Compare In re Lavinia Goodell, 39 Wis. 232, 246 (1875) (denying female applicant’s motion for admission to the bar of its court, Wisconsin Supreme Court explained: “Discussions are habitually necessary in courts of justice, which are unfit for female ears. The habitual presence of women at these would tend to relax the public sense of decency and propriety.”), with Levine, Closing Comments, 6 Law & Inequality 41 (1988) (presentation at Eighth Circuit Judicial Conference, Colorado Springs, Colo., July 17, 1987) (footnotes omitted):

“Plato questioned whether women should be afforded equal opportunity to become guardians, those elite Rulers of Platonic society. Ironically, in that most undemocratic system of government, the Republic, women’s native ability to serve as guardians was not seriously questioned. The concern was over the wrestling and exercise class in which all candidates for guardianship had to participate, for rigorous physical and mental training were prerequisites to attain the exalted status of guardian. And in accord with Greek custom, those exercise classes were conducted in the nude. Plato concluded that their virtue would clothe the women’s nakedness and that Platonic society would not thereby be deprived of the talent of qualified citizens for reasons of mere gender.”

For Plato’s full text on the equality of women, see 2 The Dialogues of Plato 302-312 (B. Jowett transl., 4th ed. 1953). Virginia, not bound to ancient Greek custom in its “rigorous physical and mental training” programs, could more readily make the accommodations necessary to draw on “the talent of [all] qualified citizens.” Cf. supra, at 550-551, n. 19.

[21] R. Morris, The Forging of the Union, 1781-1789, p. 193 (1987); see id. , at 191, setting out letter to a friend from Massachusetts patriot (later second President) John Adams, on the subject of qualifications for voting in his home State:

“[I]t is dangerous to open so fruitful a source of controversy and altercation as would be opened by attempting to alter the qualifications of voters; there will be no end of it. New claims will arise; women will demand a vote; lads from twelve to twenty-one will think their rights not enough attended to; and every man who has not a farthing, will demand an equal voice with any other, in all acts of state. It tends to confound and destroy all distinctions, and prostrate all ranks to one common level.” Letter from John Adams to James Sullivan (May 26, 1776), in 9 Works of John Adams 378 (C. Adams ed. 1854).

[*] The dissent equates our conclusion that VMI’s “asserted interest in promoting diversity” is not “`genuine,’ ” with a “charge” that the diversity rationale is “a pretext for discriminating against women.” Post, at 579-580. Of course, those are not the same thing. I do not read the Court as saying that the diversity rationale is a pretext for discrimination, and I would not endorse such a proposition. We may find that diversity was not the Commonwealth’s real reason without suggesting, or having to show, that the real reason was “antifeminism,” post, at 580. Our cases simply require that the proffered purpose for the challenged gender classification be the actual purpose, although not necessarily recorded. See ante, at 533, 535-536. The dissent also says that the interest in diversity is so transparent that having to articulate it is “absurd on its face.”Post, at 592. Apparently, that rationale was not obvious to the Mission Study Committee which failed to list it among its reasons for maintaining VMI’s all-men admissions policy.

[1] Accord, ante, at 541 (“In sum . . . , neither the goal of producing citizensoldiers, VMI’s raison d’être, nor VMI’s implementing methodology is inherently unsuitable to women” (internal quotation marks omitted; emphasis added)); ante, at 542 (“[T]he question is whether the Commonwealth can constitutionally deny to women who have the will and capacity, the training and attendant opportunities that VMI uniquely affords”); ante, at 547-548 (the “violation” is that “equal protection [has been] denied to women ready, willing, and able to benefit from educational opportunities of the kind VMI offers”);ante, at 550 (“As earlier stated, see supra, at 541-542, generalizations about `the way women are,’ estimates of what is appropriate for most women, no longer justify denying opportunity to women whose talent and capacity place them outside the average description”).

[2] This statement is supported by other evidence in the record demonstrating, by reference to both public and private institutions, that Virginia actively seeks tofoster its “`richheritage of pluralism and diversity in higher education,’ ” 1969 Report of the Virginia Commission on Constitutional Revision, quoted in relevant part at Lodged Materials 53; that Virginia views “`[o]ne special characteristic of the Virginia system [as being] its diversity,’ ” 1989 Virginia Plan for Higher Education, quoted in relevant part at Lodged Materials 64; and that in the Commonwealth’s view “[h]igher education resources should be viewed as a whole—public and private”—because “`Virginia needs the diversity inherent in a dual system of higher education,’ ” 1974 Report of the General Assembly Commission on Higher Education to the General Assembly of Virginia, quoted in 766 F. Supp. 1407, 1420 (WD Va. 1991). See also Budget Initiatives for 1990-1992 of State Council of Higher Education for Virginia 10 (June 21, 1989) (Budget Initiatives),quoted at n. 3,infra. It should be noted (for this point will be crucial to my later discussion) that these official reports quoted here, in text and footnote, regard the Commonwealth’s educational system—public and private —as a unitary one.

[3] The Commonwealth provides tuition assistance, scholarship grants, guaranteed loans, and work-study funds for residents of Virginia who attend private colleges in the Commonwealth. See, e. g., Va. Code Ann. §§ 23-38.11 to 23-38.19 (1993 and Supp. 1995) (Tuition Assistance Grant Act); §§ 23-38.30 to 23-38.44:3 (Virginia Student Assistance Authorities); Va. Code Ann. §§ 23-38.45 to 23-38.53 (1993) (College Scholarship Assistance Act); §§ 23-38.53:1 to 23-38.53:3 (Virginia Scholars Program); §§ 23-38.70, 23-38.71 (Virginia Work-Study Program). These programs involve substantial expenditures: for example, Virginia appropriated $4,413,750 (not counting federal funds it also earmarked) for the College Scholarship Assistance Program for both 1996 and 1997, and for the Tuition Assistance Grant Program appropriated $21,568,000 for 1996 and $25,842,000 for 1997. See 1996 Va. Appropriations Act, ch. 912, pt. 1, § 160.

In addition, as the parties stipulated in the District Court, the Commonwealth provides other financial support and assistance to private institutions—including single-sex colleges—through low-cost building loans, state-funded services contracts, and other programs. See, e. g., Va. Code Ann. §§ 23-30.39 to 23.30.58 (1993) (Educational Facilities Authority Act). The State Council of Higher Education for Virginia, in a 1989 document not created for purposes of this litigation but introduced into evidence, has described these various programs as a “means by which the Commonwealth can provide funding to its independent institutions, thereby helping to maintain a diverse system of higher education.” Budget Initiatives 10.

[4] The Court, unfamiliar with the Commonwealth’s policy of diverse and independent institutions, and in any event careless of state and local traditions, must be forgiven by Virginians for quoting a reference to “`the Charlottesville campus’ ” of the University of Virginia. See ante, at 538. The University of Virginia, an institution even older than VMI, though not as old as another of the Commonwealth’s universities, the College of William and Mary, occupies the portion of Charlottesville known, not as the “campus,” but as “the grounds.” More importantly, even if it were a “campus,” there would be no need to specify “the Charlottesville campus,” as one might refer to the Bloomington or Indianapolis campus of Indiana University. Unlike university systems with which the Court is perhaps more familiar, such as those in New York (e. g., the State University of New York at Binghamton or Buffalo), Illinois (University of Illinois at Urbana-Champaign or at Chicago), and California (University of California, Los Angeles, or University of California, Berkeley), there is only one University of Virginia. It happens (because Thomas Jefferson lived near there) to be located at Charlottesville. To many Virginians it is known, simply, as “the University,” which suffices to distinguish it from the Commonwealth’s other institutions offering 4-year college instruction, which include Christopher Newport College, Clinch Valley College, the College of William and Mary, George Mason University, James Madison University, Longwood College, Mary Washington University, Norfolk State University,Old Dominion University, Radford University, Virginia Commonwealth University, Virginia Polytechnic Institute and State University, Virginia State University—and, of course, VMI.

[5] The Court’s do-it-yourself approach to factfinding, which throughout is contrary to our well-settled rule that we will not “undertake to review concurrent findings of fact by two courts below in the absence of a very obvious and exceptional showing of error,” Graver Tank & Mfg. Co. v. Linde Air Products Co., 336 U. S. 271, 275 (1949) (and cases cited), is exemplified by its invocation of the experience of the federal military academies to prove that not much change would occur. See ante, at 542, n. 11; 544-545, and n. 15; 550-551, n. 19. In fact, the District Court noted that “the West Point experience” supported the theory that a coeducational VMI would have to “adopt a [different] system,” for West Point found it necessary upon becoming coeducational to “move away” from its adversative system. 766 F. Supp., at 1413, 1440. “Without a doubt . . . VMI’s present methods of training and education would have to be changed as West Point’s were.” Id., at 1413, n. 8; accord, 976 F. 2d 890, 896-897 (CA4 1992) (upholding District Court’s findings that “the unique characteristics of VMI’s program,” including its “unique methodology,” “would be destroyed by coeducation”).

[6] The Court is incorrect in suggesting that the Court of Appeals applied a “deferential” “brand of review inconsistent with the more exacting standard our precedent requires.” Ante, at 555. That court “inquir[ed] (1) whether the state’s objective is legitimate and important,’ and (2) whether `the requisite direct, substantial relationship between objective and means is present,’ ” 44 F. 3d, at 1235 (quoting Mississippi Univ. for Women v. Hogan, 458 U. S. 718, 725 (1982)). To be sure, such review is “deferential” to a degree that the Court’s new standard is not, for it is intermediate scrutiny. (The Court cannot evade this point or prove the Court of Appeals too deferential by stating that that court “devised another test, a `substantive comparability’ inquiry,’ ” ante, at 555 (quoting 44 F. 3d, at 1237), for as that court explained, its “substantive comparability” inquiry was an “additional step” that it engrafted on “th[e] traditional test” of intermediate scrutiny, ibid. (emphasis added).)

[7] The concurrence states that it “read[s] the Court” not “as saying that the diversity rationale is a pretext” for discriminating against women, but as saying merely that the diversity rationale is not genuine. Ante, at 562, n. The Court itself makes no such disclaimer, which would be difficult to credit inasmuch as the foundation for its conclusion that the diversity rationale is not “genuin[e],” ante, at 539, is its antecedent discussion of Virginia’s “deliberate” actions over the past century and a half, based on “[f]amiliar arguments,” that sought to enforce once “widely held views about women’s proper place,” ante, at 537, 538.

[8] In this regard, I note that the Court—which I concede is under no obligation to do so—provides no example of a program that would pass muster under its reasoning today: not even, for example, a football or wrestling program. On the Court’s theory, any woman ready, willing, and physically able to participate in such a program would, as a constitutional matter, be entitled to do so.