delivered the opinion of the Court.
The interaction of two sections of an Oklahoma statute, Okla. Stat., Tit. 37, §§ 241 and 245 (.1958 and Supp. 1976) ,1 *192prohibits the sale of “nonintoxicating” 3.2% beer to males under the age of 21 and to females under the age of 18. The question to be decided is whether such a gender-based differential constitutes a denial to males 18-20 years of age of the equal protection of the laws in violation of the Fourteenth Amendment.
This action was brought in the District Court for the Western District of Oklahoma on December 20, 1972, by appellant Craig, a male then between 18 and 21 years of age, and by appellant Whitener, a licensed vendor of 3.2% beer. The complaint sought declaratory and injunctive relief against enforcement of the gender-based differential on the ground that it constituted invidious discrimination against males 18-20 years of age. A three-judge court convened under 28 U. S. C. § 2281 sustained the constitutionality of the statutory differential and dismissed the action. 399 F. Supp. 1304 (1975). We noted probable jurisdiction of appellants’ appeal, 423 U. S. 1047 (1976). We reverse.
I
We first address a preliminary question of standing. Appellant Craig attained the age of 21 after we noted probable jurisdiction. Therefore, since only declaratory and injunctive relief against enforcement of the gender-based differential is sought, the controversy has been rendered moot as to Craig. See, e. g., DeFunis v. Odegaard, 416 U. S. 312 (1974).2 The question thus arises whether appellant Whitener, the licensed vendor of 3.2% beer, who has a live controversy against enforcement of the statute, may rely upon the equal protection objections of males 18-20 years of age to establish her claim of *193unconstitutionality of the age-sex differential. We conclude that she may.
Initially, it should be noted that, despite having had the opportunity to do so,3 appellees never raised before the District Court any objection to Whitener’s reliance upon the claimed unequal treatment of 18-20-year-old males as the premise of her equal protection challenge to Oklahoma’s 3.2% beer law. See 399 P. Supp., at 1306 n. 1. Indeed, at oral argument Oklahoma acknowledged that appellees always “presumed” that the vendor, subject to sanctions and loss of license for violation of the statute, was a proper party in interest to object to the enforcement of the sex-based regulatory provision. Tr. of Oral Arg. 41. While such a concession certainly would not be controlling upon the reach of this Court’s constitutional authority to exercise jurisdiction under Art. III, see, e. g., Sierra Club v. Morton, 405 U. S. 727, 732 n. 3 (1972); cf. Data Processing Service v. Camp, 397 U. S. 150, 151 (1970), our decisions have settled that limitations on a litigant’s assertion of jus tertii are not constitutionally mandated, but rather stem from a salutary “rule of self-restraint” designed to minimize unwarranted intervention into controversies where the applicable constitutional questions are ill-defined and speculative. See, e. g., Barrows v. Jackson, 346 U. S. 249, 255, 257 (1953); see also Singleton v. Wulff, 428 U. S. 106, 123-124 (1976) (Powell, J.; dissenting). These prudential objectives, thought to be enhanced by restrictions on third-party standing, cannot be furthered here, where the lower court already has entertained the relevant constitutional challenge and the parties have sought — or at least have never resisted — an authoritative constitutional determination. In such circumstances, a decision by us to forgo *194consideration of the constitutional merits in order to await the initiation of a new challenge to the statute by injured third parties would be impermissibly to foster repetitive and time-consuming litigation under the guise of caution and prudence. Moreover, insofar as the applicable constitutional questions have been and continue to be presented vigorously and “cogently,” Holden v. Hardy, 169 U. S. 366, 397 (1898), the denial of jus tertii standing in deference to a direct class suit can serve no functional purpose. Our Brother Blackmun’s comment is pertinent: “[I]t may be that a class could be assembled, whose fluid membership always included some [males] with live claims. But if the assertion of the right is to be ‘representative’ to such an extent anyway, there seems little loss in terms of effective advocacy from allowing its assertion by” the present jus tertii champion. Singleton v. Wulff, supra, at 117-118.
In any event, we conclude that appellant Whitener has established independently her claim to assert jus tertii standing. The operation of §§ 241 and 245 plainly has inflicted “injury in fact” upon appellant sufficient to guarantee her “concrete adverseness,” Baker v. Carr, 369 U. S. 186, 204 (1962), and to satisfy the constitutionally based standing requirements imposed by Art. III. The legal duties created by the statutory sections under challenge are addressed directly to vendors such as appellant. She is obliged either to heed the statutory discrimination, thereby incurring a direct economic injury through the constriction of her buyers’ market, or to disobey the statutory command and suffer, in the words of Oklahoma’s Assistant Attorney General, “sanctions and perhaps loss of license.” Tr. of Oral Arg. 41. This Court repeatedly has recognized that such injuries establish the threshold requirements of a “case or controversy” mandated by Art. III. See, e. g., Singleton v. Wulff, supra, at 113 (doctors who receive payments for their abortion services are “classically adverse” to government as payer); Sullivan v. Little Hunting *195Park, 396 U. S. 229, 237 (1969); Barrows v. Jackson, supra, at 255-256.
As a vendor with standing to challenge the lawfulness of § § 241 and 245, appellant Whitener is entitled to assert those concomitant rights of third parties that would be “diluted or adversely affected” should her constitutional challenge fail and the statutes remain in force. Griswold v. Connecticut, 381 U. S. 479, 481 (1965); see Note, Standing to Assert Constitutional Jus Tertii, 88 Harv. L. Rev. 423, 432 (1974). Otherwise, the threatened imposition of governmental sanctions might deter appellant Whitener and other similarly situated vendors from selling 3.2% beer to young males, thereby ensuring that “enforcement of the challenged restriction against the [vendor] would result indirectly in the violation of third parties' rights.” Warth v. Seldin, 422 U. S. 490, 510 (1975). Accordingly, vendors and those in like positions have been uniformly permitted to resist efforts at restricting their operations by acting as advocates of the rights of third parties who seek access to their market or function. See, e. g., Eisenstadt v. Baird, 405 U. S. 438 (1972); Sullivan v. Little Hunting Park, supra; Barrows v. Jackson, supra.4
*196Indeed, the jus tertii question raised here is answered by our disposition of a like argument in Eisenstadt v. Baird, supra. There, as here, a state statute imposed legal duties and disabilities upon the claimant, who was convicted of distributing a package of contraceptive foam to a third party.5 Since the statute was directed at Baird and penalized his conduct, the Court did not hesitate — again as here — to conclude that the “case or controversy” requirement of Art. Ill was satisfied. 405 U. S., at 443. In considering Baird’s constitutional objections, the Court fully recognized his standing to defend the privacy interests of third parties. Deemed crucial to the decision to permit jus tertii standing was the recognition of “the impact of the litigation on the third-party interests.” Id., at 445. Just as the defeat of Baird’s suit and the “[enforcement of the Massachusetts statute will materially impair the ability of single persons to obtain contraceptives,” id., at 446, so too the failure of Whitener to prevail in this suit and the continued enforcement of §§ 241 and 245 will “materially impair the ability of” males 18-20 years of age to purchase 3.2% beer despite their classification by an overt gender-based criterion. Similarly, just as the Massachusetts law in Eisenstadt “prohibit[ed], *197not use, but distribution,” 405 U. S., at 446, and consequently the least awkward challenger was one in Baird’s position who was subject to that proscription, the law challenged here explicitly regulates the sale rather than use of 3.2% beer, thus leaving a vendor as the obvious claimant.
We therefore hold that Whitener has standing to- raise relevant equal protection challenges to Oklahoma’s gender-based law. We now consider those arguments.
II
A
Before 1972, Oklahoma defined the commencement of civil majority at age 18 for females and age 21 for males. Okla. Stat., Tit. 15, § 13 (1972 and Supp. 1976). In contrast, females were held criminally responsible as adults at age 18 and males at age 16. Okla. Stat., Tit. 10, § 1101 (a) (Supp. 1976). After the Court of Appeals for the Tenth Circuit held in 1972, on the authority of Reed v. Reed, 404 U. S. 71 (1971), that the age distinction was unconstitutional for purposes of establishing criminal responsibility as adults, Lamb v. Brown, 456 F. 2d 18, the Oklahoma Legislature fixed age 18 as applicable to both males and females. Okla. Stat., Tit. 10, § 1101 (a) (Supp. 1976). In 1972, 18 also was established as the age of majority for males and females in civil matters, Okla. Stat., Tit. 15, § 13 (1972 and Supp. 1976), except that §§241 and 245 of the 3.2% beer statute were simultaneously codified to create an exception to the gender-free rule.
Analysis may appropriately begin with the reminder that Reed emphasized that statutory classifications that distinguish between males and females are “subject to scrutiny under the Equal Protection Clause.” 404 U. S., at 75. To withstand constitutional challenge, previous cases establish that classifications by gender must serve important governmental objectives and must be substantially related to achievement of those objectives. Thus, in Reed, the ob*198jectives of “reducing the workload on probate courts,” id., at 76, and “avoiding intrafamily controversy,” id., at 77, were deemed of insufficient importance to sustain use of an overt gender criterion in the appointment of administrators of intestate decedents’ estates. Decisions following Reed similarly have rejected administrative ease and convenience as sufficiently important objectives to justify gender-based classifications. See, e. g., Stanley v. Illinois, 405 U. S. 645, 656 (1972); Frontiero v. Richardson, 411 U. S. 677, 690 (1973); cf. Schlesinger v. Ballard, 419 U. S. 498, 506-507 (1975). And only two Terms ago, Stanton v. Stanton, 421 U. S. 7 (1975), expressly stating that Reed v. Reed was “controlling,” 421 U. S., at 13, held that Reed required invalidation of a Utah differential age-of-majority statute, notwithstanding the statute’s coincidence with and furtherance of the State’s purpose of fostering “old notions” of role typing and preparing boys for their expected performance in the economic and political worlds. 421 U. S., at 14-15.6
Reed v. Reed has also provided the underpinning for decisions that have invalidated statutes employing gender as an inaccurate proxy for other, more germane bases of classification. Hence, “archaic. and overbroad” generalizations, Schlesinger v. Ballard, supra, at 508, concerning the financial position of servicewomen, Frontiero v. Richardson, supra, at 689 n. 23, and working women, Weinberger v. Wiesenfeld, 420 U. S. 636, 643 (1975), could not justify use of a gender line in determining eligibility for certain governmental entitlements. Similarly, increasingly outdated *199misconceptions concerning the role of females in the home rather than in the “marketplace and world of ideas” were rejected as loose-fitting characterizations incapable of supporting state statutory schemes that were premised upon their accuracy. Stanton v. Stanton, supra; Taylor v. Louisiana, 419 U. S. 522, 535 n. 17 (1975). In light of the weak congruence between gender and the characteristic or trait that gender purported to represent, it was necessary that the legislatures choose either to realign their substantive laws in a gender-neutral fashion, or to adopt procedures for identifying those instances where the sex-centered generalization actually comported with fact. See, e. g., Stanley v. Illinois, supra, at 658; cf. Cleveland Board of Education v. LaFleur, 414 U. S. 632, 650 (1974).
In this case, too, “Reed, we feel, is controlling ...,” Stanton v. Stanton, supra, at 13. We turn then to the question whether, under Reed, the difference between males and females with respect to the purchase of 3.2% beer warrants the differential in age drawn by the Oklahoma statute. We conclude that it does not.
B
The District Court recognized that Reed v. Reed was controlling. In applying the teachings of that case, the court found the requisite important governmental objective in the traffic-safety goal proffered by the Oklahoma Attorney General, It then concluded that the statistics introduced by the appellees established that the gender-based distinction was substantially related to achievement of that goal.
C
We accept for purposes of discussion the District Court’s identification of the objective underlying §§ 241 and 245 as the enhancement of traffic safety.7 Clearly; the protection *200of public health and safety represents an important function of state and local governments However, appellees’ statistics in our view cannot support the conclusion that the gender-based distinction closely serves to achieve that objective and therefore the distinction cannot under Reed withstand equal protection challenge.
The appellees introduced a variety of statistical surveys. First, an analysis of arrest statistics for 1973 demonstrated that 18-20-year-old male arrests for “driving' under the influence” and “drunkenness” substantially exceeded female arrests for that same age period.8 Similarly, youths aged 17-21 were found to be overrepresented among those killed *201or injured in traffic accidents, with males again numerically exceeding females in this regard.9 Third, a random roadside survey in Oklahoma City revealed that young males were more inclined to drive and drink beer than were their female counterparts.10 Fourth, Federal Bureau of Investigation nationwide statistics exhibited a notable increase in arrests for “driving under the influence.”11 Finally, statistical evidence gathered in other jurisdictions, particularly Minnesota and Michigan, was offered to corroborate Oklahoma’s experience by indicating the pervasiveness of youthful participation in motor vehicle accidents following the imbibing of alcohol. Conceding that “the case is not free from doubt,” 399 F. Supp., at 1314, the District Court nonetheless concluded that this statistical showing substantiated “a rational basis for the legislative judgment underlying the challenged classification.” Id., at 1307.
Even were this statistical evidence accepted as accurate, it nevertheless offers only a weak answer to the equal protection question presented here. The most focused and relevant of the statistical surveys, arrests of 18-20-year-olds for alcohol-related driving offenses, exemplifies the ultimate unpersuasiveness of this evidentiary record. Viewed in terms of the correlation between sex and the actual activity that Oklahoma seeks to regulate — driving while under the influence of alcohol — the statistics broadly establish that .18% of females and 2% of males in that age group were arrested for that offense. While such a disparity is not trivial in a statistical sense, it hardly can form the basis for employment of a gender line as a classifying device. Certainly if male*202ness is to serve as a proxy for drinking and driving, a correlation of 2% must be considered an unduly tenuous “fit.” 12 Indeed, prior cases have consistently rejected the use of sex as a decisionmaking factor even though the statutes in question certainly rested on far more predictive empirical relationships than this.13
Moreover, the statistics exhibit a variety of other shortcomings that seriously impugn their value to equal protection analysis. Setting aside the obvious methodological problems,14 the surveys do not adequately justify the salient *203features of Oklahoma’s gender-based traffic-safety law. None purports to measure the use and dangerousness of 3.2% beer as opposed to alcohol generally, a detail that is of particular importance since, in light of its low alcohol level, Oklahoma apparently considers the 3.2% beverage to be “nonintoxicating.” Okla. Stat., Tit. 37, § 163.1 (1958); see State ex rel. Springer v. Bliss, 199 Okla. 198, 185 P. 2d 220 (1947). Moreover, many of the studies, while graphically documenting the unfortunate increase in driving while under the influence of alcohol, make no effort to relate their findings to age-sex differentials as involved here.15 Indeed, the only survey that explicitly centered its attention upon young drivers and their use of beer — albeit apparently not of the diluted 3.2% variety — reached results that hardly can be viewed as impressive in justifying either a gender or age classification.16
*204There is no reason to belabor this line of analysis. It is unrealistic to expect either members of the judiciary or state officials to be well versed in the rigors of experimental or statistical technique. But this merely illustrates that proving broad sociological propositions by statistics is a dubious business, and one that inevitably is in tension with the normative philosophy that underlies the Equal Protection Clause.17 Suffice to say that the showing offered by the appellees does not satisfy us that sex represents a legitimate, accurate proxy for the regulation of drinking and driving. In fact, when it is further recognized that Oklahoma’s statute prohibits only the selling of 3.2% beer to young males and not their drinking the beverage once acquired (even after purchase by their 18-20-year-old female companions), the relationship between gender and traffic safety becomes far too tenuous to satisfy Reed’s requirement that the gender-based difference be substantially related to achievement of the statutory objective.
We hold, therefore, that under Reed, Oklahoma’s 3.2% beer statute invidiously discriminates against males 18-20 years of age.
D
Appellees argue, however, that §§241 and 245 enforce state policies concerning the sale and distribution of alcohol and by force of the Twenty-first Amendment should therefore be held to withstand the equal protection challenge. The District Court’s response to this contention is unclear. The court assumed that the Twenty-first Amendment “strengthened” the State’s police powers with respect to alcohol regulation, 399 F. Supp., at 1307, but then said that “the standards of review that [the Equal Protection Clause] mandates are not relaxed.” Id., at 1308. Our view is, and we hold, that the Twenty-first Amendment does not save the *205invidious gender-based discrimination from invalidation as a denial of equal protection of the laws in violation of the Fourteenth Amendment.
The history of state regulation of alcoholic beverages dates from long before adoption of the Eighteenth Amendment. In the License Cases, 5 How. 504, 579 (1847), the Court recognized a broad authority in state governments to regulate the trade of alcoholic beverages within their borders free from implied restrictions under the Commerce Clause. Later in the century, however, Leisy v. Hardin, 135 U. S. 100 (1890), undercut the theoretical underpinnings of the License Cases. This led Congress, acting pursuant to its powers under the Commerce Clause, to reinvigorate the State's regulatory role through the passage of the Wilson18 and Webb-Kenyon Acts.19 See, e. g., Clark Distilling Co. v. Western Maryland R. Co., 242 U. S. 311 (1917) (upholding Webb-Kenyon Act); In re Rahrer, 140 U. S. 545 (1891) (upholding Wilson Act). With passage of the Eighteenth Amendment, the uneasy tension between the Commerce Clause and state police power temporarily subsided.
The Twenty-first Amendment repealed the Eighteenth Amendment in 1933. The wording of § 2 of the Twenty-first Amendment20 closely follows the Webb-Kenyon and Wil*206son Acts, expressing the framers’ clear intention of constitutionalizing the Commerce Clause framework established under those statutes. This Court’s decisions since have confirmed that the Amendment primarily created an exception to the normal operation of the Commerce Clause. See, e. g., Hostetter v. Idlewild Bon Voyage Liquor Corp., 377 U. S. 324, 330 (1964); Carter v. Virginia, 321 U. S. 131, 139-140 (1944) (Frankfurter, J., concurring); Finch & Co. v. McKittrick, 305 U. S. 395, 398 (1939). Even here, however, the Twenty-first Amendment does not pro tanto repeal the Commerce Clause, but merely requires that each provision “be considered in the light of the other, and in the context of the issues and interests at stake in any concrete case.” Hostetter v. Idlewild Bon Voyage Liquor Corp., supra, at 332; cf. Department of Revenue v. James Beam Distilling Co., 377 U. S. 341 (1964); Collins v. Yosemite Park & Curry Co., 304 U. S. 518 (1938).
Once passing beyond consideration of the Commerce Clause, the relevance of the Twenty-first Amendment to other constitutional provisions becomes increasingly doubtful. As one commentator has remarked: “Neither the text nor the history of the Twenty-first Amendment suggests that it qualifies individual rights protected by the Bill of Rights and the Fourteenth Amendment where the sale or use of liquor is concerned.” P. Brest, Processes of Constitutional Decision-making, Cases and Materials, 258 (1975). Any departures from this historical view have been limited and sporadic. Two States successfully relied upon the Twenty-first Amendment to respond to challenges of major liquor importers to state authority to regulate the importation and manufacture of alcoholic beverages on Commerce Clause and Fourteenth Amendment grounds. See Mahoney v. Joseph Triner Corp., 304 U. S. 401 (1938); State Board v. Young’s Market Co., *207299 U. S. 59, 64 (1936). In fact, however, the arguments in both cases centered upon importation of intoxicants, a regulatory area where the State’s authority under the Twenty-first Amendment is transparently clear, Hostetter v. Idlewild Bon Voyage Liquor Corp., supra, at 330, and n. 9, and touched upon purely economic matters that traditionally merit only the mildest review under the Fourteenth Amendment, see, e. g., Joseph E. Seagram & Sons v. Hostetter, 384 U. S. 35, 47-48, 50-51 (1966) (rejecting Fourteenth Amendment objections to state liquor laws on the strength of Ferguson v. Skrupa, 372 U. S. 726, 729-730 (1963) and Williamson v. Lee Optical Co., 348 U. S. 483 (1955)).21 Cases involving individual rights protected by the Due Process Clause have been treated in sharp contrast. For example, when an individual objected to the mandatory “posting” of her name in retail liquor establishments and her characterization as an “excessive drink [er],” the Twenty-first Amendment was held not to qualify the scope of her due process rights. Wisconsin v. Constantineau, 400 U. S. 433, 436 (1971).
It is true that California v. LaRue, 409 U. S. 109, 115 (1972), relied upon the Twenty-first Amendment to “strengthen” the State’s authority to regulate live entertainment at establishments licensed to dispense liquor, at least when the performances “partake more of gross sexuality than of communication,” id., at 118. Nevertheless, the Court has never recognized sufficient “strength” in the Amendment to defeat an otherwise established claim of invidious discrimination in violation of the Equal Protection Clause. *208Rather, Moose Lodge No. 107 v. Irvis, 407 U. S. 163, 178-179 (1972), establishes that state liquor regulatory schemes cannot work invidious discriminations that violate the Equal Protection Clause.
Following this approach, both federal and state courts uniformly have declared the unconstitutionality of gender lines that restrain the activities of customers of state-regulated liquor establishments irrespective of the operation of the Twenty-first Amendment. See, e. g., White v. Fleming, 522 F. 2d 730 (CA7 1975); Women’s Liberation Union of R. I. v. Israel, 512 F. 2d 106 (CA1 1975); Daugherty v. Daley, 370 F. Supp. 338 (ND Ill. 1974) (three-judge court) ; Seidenberg v. McSorleys’ Old Ale House, Inc., 317 F. Supp. 593 (SDNY 1970); Commonwealth Alcoholic Beverage Control Bd. v. Burke, 481 S. W. 2d 52 (Ky. 1972); cf. Sail’er Inn, Inc. v. Kirby, 5 Cal. 3d 1, 485 P. 2d 529 (1971); Paterson Tavern & G. O. A. v. Hawthorne, 57 N. J. 180, 270 A. 2d 628 (1970). Even when state officials have posited sociological or empirical justifications for these gender-based differentiations, the courts have struck down discriminations aimed at an entire class under the guise of alcohol regulation. In fact, social science studies that have uncovered quantifiable differences in drinking tendencies dividing along both racial and ethnic lines strongly suggest the need for application of the Equal Protection Clause in preventing discriminatory treatment that almost certainly would be perceived as invidious.’22 In sum, the principles embodied in the Equal *209Protection Clause are not to be rendered inapplicable by statistically measured but loose-fitting generalities concerning the drinking tendencies of aggregate- groups. We thus hold that the operation of the Twenty-first Amendment does not alter the application of equal protection standards that otherwise govern this case.
*210We conclude that the gender-based differential contained in Okla. Stat., Tit. 37, § 245 (1976 Supp.) constitutes a denial of the equal protection of the laws to males aged 18-2023 and reverse the judgment of the District Court.24
It is so ordered.
Sections 241 and 245 provide in pertinent part:
§ 241. “It shafi be unlawful for any person who holds a license to seE and dispense beer ... to sell, barter or give to any minor any beverage containing more than one-half of one per cent of alcohol measured by volume and not more than three and two-tenths (3.2) per cent of alcohol measured by weight.
§ 245. “A 'minor/ for the purposes of Section . . . 241 ... is defined as a *192female under the age of eighteen (18) years, and a male under the age of twenty-one (21) years.”
Appellants did not seek class certification of Craig as representative of other similarly situated males 18-20 years of age. See, e. g., Sosna v. Iowa, 419 U. S. 393, 401 (1975).
The District Court’s opinion confirms that Whitener from the outset has based her constitutional challenge on gender-discrimination grounds, 399 F. Supp., at 1306, and “[n]o challenge is made to [her] standing and requisite interest in the controversy . . . .” Id., at 1306 n. 1.
The standing question presented here is not answered by the principle stated in United States v. Raines, 362 U. S. 17, 21 (1960), that “one to whom application of a statute is constitutional will not be heard to attack the statute on the ground that impliedly it might also be taken as applying to other persons or other situations in which its application might be unconstitutional.” In Raines, the Court refused to permit certain public officials of Georgia to defend against application of the Civil Rights Act to their official conduct on the ground that the statute also might be construed to encompass the “purely private actions” of others. The Raines rule remains germane in such a setting, where the interests of the litigant and the rights of the proposed third parties are in no way mutually interdependent. Thus, a successful suit against Raines did not threaten to impair or diminish the independent private rights of others, and consequently, consideration of those third-party rights properly was deferred until another day.
Of course, the Raines principle has also been relaxed where legal action *196against the claimant threatens to “chill” the First Amendment rights of third parties. See, e. g., Lewis v. New Orleans, 415 U. S. 130 (1974).
The fact that Baird chose to disobey the legal duty imposed upon him by the Massachusetts anticontraception statute, resulting in his criminal conviction, 405 U. S., at 440, does not distinguish the standing inquiry from that pertaining to the anticipatory attack in this case. In both Eisenstadt and here, the challenged statutes compel jus tertii claimants either to cease their proscribed activities or to suffer appropriate sanctions. The existence of Art.. Ill “injury in fact” and the structure of the claimant’s relationship to the third parties are not altered by the litigative posture of the suit. And, certainly, no suggestion will be heard that Whitener’s anticipatory challenge offends the normal requirements governing such actions. See generally Steffel v. Thompson, 415 U. S. 452 (1974); Samuels v. Mackell, 401 U. S. 66 (1971); Younger v. Harris, 401 U.S. 37 (1971).
Kahn v. Shevin, 416 U. S. 351 (1974) and Schlesinger v. Ballard, 419 U. S. 498 (1975), upholding the use of gender-based classifications, rested upon the Court’s perception of the laudatory purposes of those laws as remedying disadvantageous conditions suffered by women in economic and military life. See 416 U. S., at 353-354; 419 U. S., at 508. Needless to say, in this case Oklahoma does not suggest that the age-sex differential was enacted to ensure the availability of 3.2% beer for women as compensation for previous deprivations.
That this was the true purpose is not at all self-evident. The pur*200pose is not apparent from the face of the statute and the Oklahoma Legislature does not preserve statutory history materials capable of clarifying the objectives served by its legislative enactments. The District Court acknowledged the nonexistence of materials necessary “to reveal what the actual purpose of the legislature was,” but concluded that “we feel it apparent that a major purpose of the legislature was to promote the safety of the young persons affected and the public generally.” 399 F. Supp., at 1311 n. 6. Similarly, the attorney for Oklahoma, while proposing traffic safety as a legitimate rationale for the 3.2% beer law, candidly acknowledged at oral argument that he is unable to assert that traffic safety is “indeed the reason" for the gender line contained in § 245. Tr. of Oral Arg. 27. For this appeal we find adequate the appellee’s representation of legislative purpose, leaving for another day consideration of whether the statement of the State’s Assistant Attorney General should suffice to inform this Court of the legislature’s objectives, or whether the Court must determine if the litigant simply is selecting a convenient, but false, post hoc rationalization.
The disparities in 18-20-year-old male-female arrests were substantial for both categories of offenses: 427 versus 24 for driving under the influence of alcohol, and 966 versus 102 for drunkenness. Even if we assume that a legislature may rely on such arrest data in some situations, these figures do not offer support for a differential age line, for the disproportionate arrests of males persisted at older ages; indeed, in the case of arrests for drunkenness, the figures for all ages indicated “even more male involvement in such arrests at later ages.” 399 F. Supp., at 1309. See also n. 14, infra.
This survey drew no correlation between the accident figures for any age group and levels of intoxication found in those killed or injured.
For an analysis of the results of this exhibit, see n. 16, infra.
The FBI made no attempt to relate these arrest figures either to beer drinking or to an 18-21 age differential, but rather found that male arrests for all ages exceeded 90% of the total.
Obviously, arrest statistics do not embrace all individuals who drink and drive. But for purposes of analysis, this "underinclusiveness” must be discounted somewhat by the shortcomings inherent in this statistical sample, see n. 14, infra. In any event, we decide this case in light of the evidence offered by Oklahoma and know of no way of extrapolating these arrest statistics to take into account the driving and drinking population at large, including those who avoided arrest.
For example, we can conjecture that in Reed, Idaho’s apparent premise that women lacked experience in formal business matters (particularly compared to men) would have proved to be accurate in substantially more than 2% of all cases. And in both Frontiero and Wiesenfeld, we expressly found appellees’ empirical defense of mandatory dependency tests for men but not women to be unsatisfactory, even though we recognized that husbands are still far less likely to be dependent on their wives than vice versa. See, e. g., 411 U. S., at 688-690.
The very social stereotypes that find reflection in age-differential laws, see Stanton v. Stanton, 421 U. S., 7, 14-15 (1975), are likely substantially to distort the accuracy of these comparative statistics. Hence “reckless” young men who drink and drive are transformed into arrest statistics, whereas their female counterparts are chivalrously escorted home. See, e. g., W. Reckless & B. Kay, The Female Offender 4, 7, 13, 16-17 (Report to Presidential Commission on Law Enforcement and Administration of Justice, 1967). Moreover, the Oklahoma surveys, gathered under a regime where the age-differential law in question has been in effect, are lacking in controls necessary for appraisal of the actual effectiveness of the male 3.2% beer prohibition. In this regard, the disproportionately high arrest statistics for young males — and, indeed, the growing alcohol-related arrest figures for all ages and sexes — simply may be taken to document the relative futility of controlling driving behavior by the 3.2% beer *203statute and like legislation, although we obviously have no means of estimating how many individuals, if any, actually were prevented from drinking by these laws.
See, e. g., nn. 9 and 11, supra. See also n. 16, infra.
The random roadside survey of drivers conducted in Oklahoma City during August 1972 found that 78% of drivers under 20 were male. Turning to an evaluation of-their drinking habits and factoring out nondrinkers, 84% of the males versus 77% of the females expressed a preference for beer. Further 16.5% of the men and 11.4% of the women had consumed some alcoholic beverage within two hours of the interview. Finally, a blood alcohol concentration greater than .01% was discovered in 14.6% of the males compared to 11.5% of the females. “The 1973 figures, although they contain some variations, reflect essentially the same pattern.” 399 F. Supp., at 1309. Plainly these statistical disparities between the sexes are not substantial. Moreover, when the 18-20 age boundaries are lifted and all drivers analyzed, the 1972 roadside survey indicates that male drinking rose slightly whereas female exposure to alcohol remained relatively constant. Again, in 1973, the survey established that “compared to all drivers interviewed, . . . the under-20 age group generally showed a lower involvement with alcohol in terms of having drunk within the past two hours or having a significant BAC (blood alcohol content).” Ibid. In sum, this survey provides little support for a gender line among teenagers and actually runs counter to the imposition of drinking restrictions based upon age.
See, e. g., n. 22, infra.
The Wilson Act, enacted in 1890, reads in pertinent part: “All . . . intoxicating liquors or liquids transported into any State or Territory . . . shall upon arrival in such State or Territory be subject to the operation and effect of the laws of such State or Territory enacted in the exercise of its police powers, to the same extent and in the same manner as though such liquids or liquors had been produced in such State or Territory . . . .” 27 U. S. C. § 121.
The Webb-Kenyon Act of 1913 prohibits “[t]he shipment or transportation ... of any . . . intoxicating liquor of any kind, from one State, Territory, or District . . . into any other State, Territory, or District . . . [for the purpose of being] received, possessed, sold, or in any manner used ... in violation of any law of such State, Territory, or District . . . .” 27 U. S. C. § 122.
“The transportation or importation into any State, Territory, or *206possession of the United States for delivery or use therein of intoxicating liquors, in violation of the laws thereof, is hereby prohibited.”
The dictum contained in State Board v. Young’s Market Co., 299 U. S. 59, 64 (1936), that “[a] classification recognized by the Twenty-first Amendment cannot be deemed forbidden by the Fourteenth,’' is inapplicable to this case. The Twenty-first Amendment does not recognize, even indirectly, classifications based upon gender. And, as the accompanying text demonstrates, that statement has not been relied upon in recent cases that have considered Fourteenth Amendment challenges to state liquor regulation.
Thus, if statistics were to govern the permissibility of state alcohol regulation without regard to the Equal Protection Clause as a limiting principle, it might follow that States could freely favor Jews and Italian Catholics at the expense of all other Americans, since available studies regularly demonstrate that the former two groups exhibit the lowest rates of problem drinking. See, e. g., Haberman & Sheinberg, Implicative Drinking Reported in a Household Survey: A Corroborative Note on Subgroup Differences, 28 Q. J. Studies on Alcohol 538 (1967); Wechsler, Thum, Demone, & Dwinnell, Social Characteristics and Blood Alcohol Level, 33 Q. J. Studies on Alcohol 132, 141-142 (1972); Wechsler, *209Demone, Thum, & Kasey, Religious-Ethnic Differences In Alcohol Consumption, 11 J. Health & Soc. Behavior 21, 28 (1970); Schmidt & Popham, Impressions of Jewish Alcoholics, 37 J. Studies on Alcohol 931 (1976). Similarly, if a State were allowed simply to depend upon demographic characteristics of adolescents in identifying problem drinkers, statistics might support the conclusion that only black teenagers should be permitted to drink, followed by Asian-Americans and Spanish-Americans. “Whites and American Indians have the lowest proportions of abstainers and the highest proportions of moderate/heavy and heavy drinkers.” Summary of Final Report of a National Study of Adolescent Drinking Behavior, Attitudes and Correlates 147-148 (Center for the Study of Social Behavior, Research Triangle Inst., Apr. 1975) (percentage of moderate/ heavy and heavy adolescent drinkers by race: black 15.2%; AsianAmeriean 18.3%; Spanish-American 22.7%; white 25.3%; American Indian 28.1%).
In the past, some States have acted upon their notions of the drinking propensities of entire groups in fashioning their. alcohol policies. The most typical recipient of this treatment has been the American Indian; indeed, several States established criminal sanctions for the sale of alcohol to an Indian or “half or quarter breed Indian.” See, e. g., Fla. Stat. Ann. § 569.07 (1962 and 1976 Supp.) (repealed in 1972); Iowa Code Ann-. §732.5 (1950 and 1976 Supp.) (repealed in 1967); Minn. Stat. Ann. § 340.82 (1957) (repealed in 1969); Neb. Rev. Stat. 53-181 (1944) (repealed in 1955); Utah Code Ann. §76-34^1 (1953 and 1975 Supp.) (repealed in 1955). Other statutes and constitutional provisions proscribed the introduction of alcoholic beverages onto Indian reservations. See, e. g., Act of June 10, 1910, §2, 36 Stat. 558; Ariz. Const., Art. XX, §3; N. M. Const., Art. XXI, §8; Okla. Const., Art. 1, §7. While Indian-oriented provisions were the most common, state alcohol beverage prohibitions also have been directed at other groups, notably German, Italian, and Catholic immigrants. See, e. g., J. Higham, Strangers in the Land 25, 267-268, 295 (1975). The repeal of most of these laws signals society’s perception of the unfairness and questionable constitutionality of singling out groups to bear the brunt of alcohol regulation.
Insofar as Goesaert v. Cleary, 335 U. S. 464 (1948), may be inconsistent, that decision is disapproved. Undoubtedly reflecting the view that Goesaert’s equal protection analysis no longer obtains, the District Court made no reference to that decision in upholding Oklahoma’s-statute. Similarly, the opinions of the federal and state courts cited earlier in the text invalidating gender lines with respect to alcohol regulation uniformly disparaged the contemporary vitality of Goesaert.
As noted in Stanton v. Stanton, 421 U. S., at 17-18, the Oklahoma Legislature is free to redefine any cutoff age for the purchase and sale of 3.2% beer that it may choose, provided that the redefinition operates in a gender-neutral fashion.