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. Closed-minded they were — as every age is, including our own, with regard to matters it cannot guess, because it simply does not *567consider 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
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. *568v. 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., *569dissenting). 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 *570today’s decision with the cutoff of all state and federal support. Government-run woranilitary 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.
II
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 “[between 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 “sub*571stantially 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 ivhether a “justification” is “exceedingly persuasive” must be assessed by asking *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 inherently 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
*573Only 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 *574objective. There is simply 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 sex-based 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 applica*575ble 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 minor-it[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; *576Violence Against Women Act of 1994, Pub. L. 103-322, Title IV, 108 Stat. 1902.
III
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.”
A
It is beyond question that Virginia has an important state interest in providing effective college education for its citizens. 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 deep-seated.” 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 pro*577vide 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 YMI 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 “ad-versative, 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 *578practice 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., adversa-tive] 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 adversa-tive 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 “ ‘[hjigher education resources should be viewed as a whole — public and pri*579vate’” — 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 adversa-tive 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.
B
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 *580charge, 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” *581is 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 “[bjefore 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.’” Stipula*582tions 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.
*583The 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 deci-*584sionmaking 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
*5853. 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 ease ... is virtually uncon-tradicted,” 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 *586who 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 “training] 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 *587Visitors, 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 fashion — i. 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 dis*588cuss, 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 *589admission 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 “[ejvidence supports this theory,” but specifically concluded that while “[without 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 adver-sative 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
*5906. 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 VMFs 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 sub*591stantially 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 adver-sative 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 *592these professionals acted on “‘overbroad’ generalizations,” ante, at 542, 550.
C
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 dis*593cussing 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 *594soon 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 *595ought 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 reason Virginia’s assistance to its four all-women’s private colleges does not count is that “[tjhe 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.
IV
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.
A
Under the constitutional principles announced and applied today, single-sex public education is unconstitutional. By going through the motions of applying a balancing test — ask*596ing 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 single-sex 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 sex-based 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. *597The 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 single-sex 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 *598Detroit 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.
B
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 *599the Court will disclaim the reasoning it has used today to destroy VML 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 unconstitu*600tional, it stated that the implications will not be so disastrous, since States can provide funding to racially segregated private schools, “depending] on the circumstances,” Tr. of Oral Arg. 56. I cannot imagine what those “circumstances” might be, and it would be as foolish for private-school 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 “unequivocal]” 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 all-male institution, conceived in chauvinism, etc., etc. I will not join that opinion.
* * *
Justice Brandéis 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(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 stu*602dents (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 peop’e with courtesy, no matter what their social position may be.
*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.
Accord, ante, at 541 (“In sum..., neither the goal of producing citizen-soldiers, VMI’s raison d’etre, 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, gen*573eralizations 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”).
This statement is supported by other evidence in the record demonstrating, by reference to both public and private institutions, that Virginia actively seeks to foster its ‘“rich heritage of pluralism and diversity in higher education,”’ 1969 Report of the Virginia Commission on Constitutional Revision, quoted in relevant part at Lodged Materials 58; 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.
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.
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, *585simply, 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.
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”).
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).)
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 “[fjamiliar arguments,” that sought to enforce once “widely held views about women’s proper place,” ante, at 537, 538.
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.