dissenting.
The Court today, for the first time, approves direct funding of core religious activities by an arm of the State. It does so, however, only after erroneous treatment of some familiar principles of law implementing the First Amendment’s Establishment and Speech Clauses, and by viewing the very funds in question as beyond the reach of the Establishment Clause’s funding restrictions as such. Because there is no *864warrant for extinguishing among public funding sources for purposes of applying the First Amendment’s prohibition of religious establishment, I would hold that the University’s refusal to support petitioners’ religious activities is compelled by the Establishment Clause. I would therefore affirm.
HH
The central question in this case is whether a grant from the Student Activities Fund to pay Wide Awake’s printing expenses would violate the Establishment Clause. Although the Court does not dwell on the details of Wide Awake’s message, it recognizes something sufficiently religious in the publication to demand Establishment Clause scrutiny. Although the Court places great stress on the eligibility of secular as well as religious activities for grants from the Student Activities Fund, it recognizes that such evenhanded availability is not by itself enough to satisfy constitutional requirements for any aid scheme that results in a benefit to religion. Ante, at 839; see also ante, at 846-848 (O’Connor, J., concurring). Something more is necessary to justify any religious aid. Some Members of the Court, at least, may think the funding permissible on a view that it is indirect, since the money goes to Wide Awake’s printer, not through Wide Awake’s own checking account. The Court’s principal reliance, however, is on an argument that providing religion with economically valuable services is permissible on the theory that services are economically indistinguishable from religious access to governmental speech forums, which sometimes is permissible. But this reasoning would commit the Court to approving direct religious aid beyond anything justifiable for the sake of access to speaking forums. The Court implicitly recognizes this in its further attempt to circumvent the clear bar to direct governmental aid to religion. Different Members of the Court seek to avoid this bar in different ways. The opinion of the Court makes the novel assumption that only direct aid financed with tax *865revenue is barred, and draws the erroneous conclusion that the involuntary Student Activities Fee is not a tax. I do not read Justice O’Connor’s opinion as sharing that assumption; she places this Student Activities Fund in a category of student binding enterprises from which religious activities in public universities may benefit, so long as there is no consequent endorsement of religion. The resulting decision is in unmistakable tension with the accepted law that the Court continues to avow.
A
The Court’s difficulties will be all the more clear after a closer look at Wide Awake than the majority opinion affords. The character of the magazine is candidly disclosed on the opening page of the first issue, where the editor-in-chief announces Wide Awake’s mission in a letter to the readership signed, “Love in Christ”: it is “to challenge Christians to live, in word and deed, according to the faith they proclaim and to encourage students to consider what a personal relationship with Jesus Christ means.” App. 45. The masthead of every issue bears St. Paul’s exhortation, that “[t]he hour has come for you to awake from your slumber, because our salvation is nearer now than when we first believed. Romans 13:11.”
Each issue of Wide Awake contained in the record makes good on the editor’s promise and echoes the Apostle’s call to accept salvation:
“The only way to salvation through Him is by confessing and repenting of sin. It is the Christian’s duty to make sinners aware of their need for salvation. Thus, Christians must confront and condemn sin, or else they fail in their duty of love.” Mourad & Prince, A Love/Hate Relationship, Nov./Dec. 1990, p. 3.
“When you get to the final gate, the Lord will be handing out boarding passes, and He will examine your ticket. If, in your lifetime, you did not request a seat *866on His Friendly Skies Flyer by trusting Him and asking Him to be your pilot, then you will not be on His list of reserved seats (and the Lord will know you not). You will not be able to buy a ticket then; no amount of money or desire will do the trick. You will be met by your chosen pilot and flown straight to Hell on an express jet (without air conditioning or toilets, of course).” Ace, The Plane Truth, ibid.
“ ‘Go into all the world and preach the good news to all creation.’ (Mark 16:15) The Great Commission is the prime-directive for our lives as Christians . . . .” Liu, Christianity and the Five-legged Stool, Sept./Oct. 1991, p. 3.
“The Spirit provides access to an intimate relationship with the Lord of the Universe, awakens our minds to comprehend spiritual truth and empowers us to serve as effective ambassadors for the Lord Jesus in our earthly lives.” Buterbaugh, A Spiritual Advantage, Mar./Apr. 1991, p. 21.
There is no need to quote further from articles of like tenor, but one could examine such other examples as religious poetry, see Macpherson, I Have Started Searching for Angels, Nov. /Dec. 1990, p. 18; religious textual analysis and commentary, see Buterbaugh, Colossians 1:1-14: Abundant Life, id., at 20; Buterbaugh, John 14-16: A Spiritual Advantage, Mar./Apr., pp. 20-21; and instruction on religious practice, see Early, Thanksgiving and Prayer, Nov./Dec. 1990, p. 21 (providing readers with suggested prayers and posing contemplative questions about biblical texts); Early, Hope and Spirit, Mar./Apr. 1991, p. 21 (similar).
Even featured essays on facially secular topics become platforms from which to call readers to fulfill the tenets of Christianity in their lives. Although a piece on racism has some general discussion on the subject, it proceeds beyond even the analysis and interpretation of biblical texts to con-*867elude with the counsel to take action because that is the Christian thing to do:
“God calls us to take the risks of voluntarily stepping out of our comfort zones and to take joy in the whole richness of our inheritance in the body of Christ. We must take the love we receive from God and share it with all peoples of the world.
“Racism is a disease of the heart, soul, and mind, and only when it is extirpated from the individual consciousness and replaced with the love and peace of God will true personal and communal healing begin.” Liu, Rosenberger, Mourad, and Prince, “Eracing” Mistakes, Nov./Dee. 1990, p. 14.
The same progression occurs in an article on eating disorders, which begins with descriptions of anorexia and bulimia and ends with this religious message:
“As thinking people who profess a belief in God, we must grasp firmly the truth, the reality of who we are because of Christ. Christ is the Bread of Life (John 6:35). Through Him, we are full. He alone can provide the ultimate source of spiritual fulfillment which permeates the emotional, psychological, and physical dimensions of our lives.” Ferguson & Lassiter, From Calorie to Calvary, Sept. /Oct. 1991, p. 14.
This writing is no merely descriptive examination of religious doctrine or even of ideal Christian practice in confronting life’s social and personal problems. Nor is it merely the expression of editorial opinion that incidentally coincides with Christian ethics and reflects a Christian view of human obligation. It is straightforward exhortation to enter into a relationship with God as revealed in Jesus Christ, and to satisfy a series of moral obligations derived from the teachings of Jesus Christ. These are not the words of “student news, information, opinion, entertainment, or academic com-municatio[n]...” (in the language of the University’s funding *868criterion, App. to Pet. for Cert. 61a), but the words of “challenge [to] Christians to live, in word and deed, according to the faith they proclaim and ... to consider what a personal relationship with Jesus Christ means” (in the language of Wide Awake’s founder, App. 46). The subject is not the discourse of the scholar’s study or the seminar room, but of the evangelist’s mission station and the pulpit. It is nothing other than the preaching of the word, which (along with the sacraments) is what most branches of Christianity offer those called to the religious life.
Using public funds for the direct subsidization of preaching the word is categorically forbidden under the Establishment Clause, and if the Clause was meant to accomplish nothing else, it was meant to bar this use of public money. Evidence on the subject antedates even the Bill of Rights itself, as may be seen in the writings of Madison, whose authority on questions about the meaning of the Establishment Clause is well settled, e. g., Committee for Public Ed. & Religious Liberty v. Nyquist, 413 U. S. 756, 770, n. 28 (1973); Everson v. Board of Ed. of Ewing, 330 U. S. 1, 13 (1947). Four years before the First Congress proposed the First Amendment, Madison gave his opinion on the legitimacy of using public funds for religious purposes, in the Memorial and Remonstrance Against Religious Assessments, which played the central role in ensuring the defeat of the Virginia tax assessment bill in 1786 and framed the debate upon which the Religion Clauses stand:
“Who does not see that . . . the same authority which can force a citizen to contribute three pence only of his property for the support of any one establishment, may force him to conform to any other establishment in all cases whatsoever?” James Madison, Memorial and Remonstrance Against Religious Assessments ¶ 3 (hereinafter Madison’s Remonstrance), reprinted in Everson, supra, at 65-66 (appendix to dissent of Rutledge, J.).
*869Madison wrote against a background in which nearly every Colony had exacted a tax for church support, Everson, supra, at 10, n. 8, the practice having become “so commonplace as to shock the freedom-loving colonials into a feeling of abhorrence,” 330 U. S., at 11 (footnote omitted). Madison’s Remonstrance captured the colonists’ “conviction that individual religious liberty could be achieved best under a government which was stripped of all power to tax, to support, or otherwise to assist any or all religions, or to interfere with the beliefs of any religious individual or group.” Ibid.1 Their sentiment, as expressed by Madison in Virginia, *870led not only to the defeat of Virginia’s tax assessment bill, but also directly to passage of the Virginia Bill for Establishing Religious Freedom, written by Thomas Jefferson. That *871bill’s preamble declared that “to compel a man to furnish contributions of money for the propagation of opinions which he disbelieves, is sinful and tyrannical,” Jefferson, A Bill for Establishing Religious Freedom, reprinted in 5 The Founder’s Constitution 84 (P. Kurland & R. Lerner eds. 1987), and its text provided “[t]hat no man shall be compelled to frequent or support any religious worship, place, or ministry whatsoever . . . id., at 85. See generally Everson, 330 U. S., at 13. We have “previously recognized that the provisions of the First Amendment, in the drafting and adoption of which Madison and Jefferson played such leading roles, had the same objective and were intended to provide the same protection against governmental intrusion on religious liberty as the Virginia statute.” Ibid.; see also Laycock, “Nonpreferential” Aid to Religion: A False Claim About Original Intent, 27 Wm. & Mary L. Rev. 875, 921, 923 (1986) (“[I]f the debates of the 1780’s support any proposition, it is that the Framers opposed government financial support for religion. . . . They did not substitute small taxes for large taxes; three pence was as bad as any larger sum. The principle was what mattered. With respect to money, religion was to be wholly voluntary. Churches either would support *872themselves or they would not, but the government would neither help nor interfere”) (footnote omitted); T. Curry, The First Freedoms 217 (1986) (At the time of the framing of the Bill of Rights, “[t]he belief that government assistance to religion, especially in the form of taxes, violated religious liberty had a long history”); J. Choper, Securing Religious Liberty 16 (1995) (“There is broad consensus that a central threat to the religious freedom of individuals and groups— indeed, in the judgment of many the most serious infringement upon religious liberty — is posed by forcing them to pay taxes in support of a religious establishment or religious activities”) (footnotes omitted; internal quotation marks omitted).2
*873The principle against direct funding with public money is patently violated by the contested use of today’s student activity fee.3 Like today’s taxes generally, the fee is Madison’s threepence. The University exercises the power of the State to compel a student to pay it, see Jefferson’s Preamble, supra, and the use of any part of it for the direct support of religious activity thus strikes at what we have repeatedly *874held to be the heart of the prohibition on establishment. Everson, 330 U. S., at 15-16 (“The ‘establishment of religion’ clause . . . means at least this .... No tax in any amount, large or small, can be levied to support any religious activities or institutions, whatever they may be called, or whatever form they may adopt to teach or practice religion”); see School Dist. of Grand Rapids v. Ball, 473 U. S. 373, 385 (1985) (“Although Establishment Clause jurisprudence is characterized by few absolutes, the Clause does absolutely prohibit government-financed or government-sponsored indoctrination into the beliefs of a particular religious faith”); Committee for Public Ed. v. Nyquist, 413 U. S., at 780 (“In the absence of an effective means of guaranteeing that the state aid derived from public funds will be used exclusively for secular, neutral, and nonideological purposes, it is clear from our cases that direct aid in whatever form is invalid”); id., at 772 (“Primary among those evils” against which the Establishment Clause guards “have been sponsorship, financial support, and active involvement of the sovereign in religious activity”) (citations and internal quotation marks omitted); see also Lee v. Weisman, 505 U. S. 577, 640 (1992) (Scalia, J., dissenting) (“The coercion that was a hallmark of historical establishments of religion was coercion of religious orthodoxy and of financial support by force of law and threat of penalty”) (emphasis deleted); cf. Flast v. Cohen, 392 U. S. 83, 103-104 (1968) (holding that taxpayers have an adequate stake in the outcome of Establishment Clause litigation to satisfy Article III standing requirements, after stating that “[o]ur history vividly illustrates that one of the specific evils feared by those who drafted the Establishment Clause and fought for its adoption was that the taxing and spending power would be used to favor one religion over another or to support religion in general”).
The Court, accordingly, has never before upheld direct state funding of the sort of proselytizing published in Wide *875Awake and, in fact, has categorically condemned state programs directly aiding religious activity, School Dist. v. Ball, supra, at 395 (striking programs providing secular instruction to nonpublic school students on nonpublic school premises because they are “indistinguishable from the provision of a direct cash subsidy to the religious school that is most clearly prohibited under the Establishment Clause”); Wolman v. Walter, 433 U. S. 229, 254 (1977) (striking field trip aid program because it constituted “an impermissible direct aid to sectarian education”); Meek v. Pittenger, 421 U. S. 349, 365 (1975) (striking material and equipment loan program to nonpublic schools because of the inability to “channe[l] aid to the secular without providing direct aid to the sectarian”); Committee for Public Ed. v. Nyquist, supra, at 774 (striking aid to nonpublic schools for maintenance and repair of facilities because “[n]o attempt is made to restrict payments to those expenditures related to the upkeep of facilities used exclusively for secular purposes”); Levitt v. Committee for Public Ed. & Religious Liberty, 413 U. S. 472, 480 (1973) (striking aid to nonpublic schools for state-mandated tests because the State had failed to “assure that the state-supported activity is not being used for religious indoctrination”); Tilton v. Richardson, 403 U. S. 672, 683 (1971) (plurality opinion) (striking as insufficient a 20-year limit on prohibition for religious use in federal construction program for university facilities because unrestricted use even after 20 years “is in effect a contribution of some value to a religious body”); id., at 689 (Douglas, J., joined by Black, and Marshall, JJ., concurring in part and dissenting in part).
Even when the Court has upheld aid to an institution performing both secular and sectarian functions, it has always made a searching enquiry to ensure that the institution kept the secular activities separate from its sectarian ones, with any direct aid flowing only to the former and never the latter. Bowen v. Kendrick, 487 U. S. 589, 614-615 (1988) (upholding *876grant program for services related to premarital adolescent sexual relations on ground that funds cannot be “used by the grantees in such a way as to advance religion”); Roemer v. Board of Public Works of Md., 426 U. S. 736, 746-748, 755, 759-761 (1976) (plurality opinion) (upholding general aid program restricting uses of funds to secular activities only); Hunt v. McNair, 413 U. S. 734, 742-745 (1973) (upholding general revenue bond program excluding from participation facilities used for religious purposes); Tilton v. Richardson, supra, at 679-682 (plurality opinion) (upholding general aid program for construction of academic facilities as “[t]here is no evidence that religion seeps into the use of any of these facilities”); see Board of Ed. of Central School Dist. No. 1 v. Allen, 392 U. S. 236, 244-248 (1968) (upholding textbook loan program limited to secular books requested by individual students for secular educational purposes).
Reasonable minds may differ over whether the Court reached the correct result in each of these cases, but their common principle has never been questioned or repudiated. “Although Establishment Clause jurisprudence is characterized by few absolutes, the Clause does absolutely prohibit government-financed ... indoctrination into the beliefs of a particular religious faith.” School Dist. v. Ball, 473 U. S., at 385.
B
Why does the Court not apply this clear law to these clear facts and conclude, as I do, that the funding scheme here is a clear constitutional violation? The answer must be in part that the Court fails to confront the evidence set out in the preceding section. Throughout its opinion, the Court refers uninformatively to Wide Awake’s “Christian viewpoint,” ante, at 826, or its “religious perspective,” ante, at 832, and in distinguishing funding of Wide Awake from the funding of a church, the Court maintains that “[Wide Awake] is not a religious institution, at least in the usual sense,” ante, at *877844;4 see also ante, at 826. The Court does not quote the magazine’s adoption of Saint Paul’s exhortation to awaken to the nearness of salvation, or any of its articles enjoining readers to accept Jesus Christ, or the religious verses, or the religious textual analyses, or the suggested prayers. And so it is easy for the Court to lose sight of what the University students and the Court of Appeals found so obvious, and to blanch the patently and frankly evangelistic character of the magazine by unrevealing allusions to religious points of view.
Neverthéless, even without the encumbrance of detail from Wide Awake’s actual pages, the Court finds something sufficiently religious about the magazine to require examination under the Establishment Clause, and one may therefore ask why the unequivocal prohibition on direct funding does not lead the Court to conclude that funding would be unconstitutional. The answer is that the Court focuses on a subsidiary body of law, which it correctly states but ultimately misapplies. That subsidiary body of law accounts for the Court’s substantial attention to the fact that the University’s funding scheme is “neutral,” in the formal sense that it makes funds available on an evenhanded basis to secular and sectarian applicants alike. Ante, at 839-842. While this is indeed true and relevant under our cases, it does not alone satisfy the requirements of the Establishment Clause, as the Court recognizes when it says that evenhandedness is only a “significant factor” in certain Establishment Clause analysis, not a dispositive one. Ante, at 839; see ante, at 840-841; see also ante, at 846-848 (O’Connor, J., concurring); ante, at 846 (“Neutrality, in both form and effect, is one hallmark of the Establishment Clause”); Capitol Square Review and Advisory Bd. v. Pinette, ante, at 777 (O’Connor, J., concur*878ring in part and concurring in judgment) (“[T]he Establishment Clause forbids a State to hide behind the application of formally neutral criteria and remain studiously oblivious to the effects of its actions. . . . [N]ot all state policies are permissible under the Religion Clauses simply because they are neutral in form”). This recognition reflects the Court’s appreciation of two general rules: that whenever affirmative government aid ultimately benefits religion, the Establishment Clause requires some justification beyond evenhandedness on the government’s part; and that direct public funding of core sectarian activities, even if accomplished pursuant to an evenhanded program, would be entirely inconsistent with the Establishment Clause and would strike at the very heart of the Clause’s protection. See ante, at 842 (“We do not confront a case where, even under a neutral program that includes nonsectarian recipients, the government is making direct money payments to an institution or group that is engaged in religious activity”); ante, at 840-841, 844; see also ante, at 847 (O’Connor, J., concurring) (“[Our] decisions ... provide no precedent for the use of public funds to finance religious activities”).
In order to understand how the Court thus begins with sound rules but ends with an unsound result, it is necessary to explore those rules in greater detail than the Court does. As the foregoing quotations from the Court’s opinion indicate, the relationship between the prohibition on direct aid and the requirement of evenhandedness when affirmative government aid does result in some benefit to religion reflects the relationship between basic rule and marginal criterion. At the heart of the Establishment Clause stands the prohibition against direct public funding, but that prohibition does not answer the questions that occur at the margins of the Clause’s application. Is any government activity that provides any incidental benefit to religion likewise unconstitutional? Would it be wrong to put out fires in burning churches, wrong to pay the bus fares of students on the way *879to parochial schools, wrong to allow a grantee of special education funds to spend them at a religious college? These are the questions that call for drawing lines, and it is in drawing them that evenhandedness becomes important. However the Court may in the past have phrased its line-drawing test, the question whether such benefits are provided on an evenhanded basis has been relevant, for the question addresses one aspect of the issue whether a law is truly neutral with respect to religion (that is, whether the law either “advance[s] [or] inhibits] religion,” County of Allegheny v. American Civil Liberties Union, Greater Pittsburgh Chapter, 492 U. S. 573, 592 (1989)). In Widmar v. Vincent, 454 U. S. 263, 274 (1981), for example, we noted that “[t]he provision of benefits to [a] broad . . . spectrum of [religious and nonreligious] groups is an important index of secular effect.” See also Board of Ed. of Kiryas Joel Village School Dist. v. Grumet, 512 U. S. 687, 702-705 (1994). In the doubtful cases (those not involving direct public funding), where there is initially room for argument about a law’s effect, evenhandedness serves to weed out those laws that impermissibly advance religion by channelling aid to it exclusively. Even-handedness is therefore a prerequisite to further enquiry into the constitutionality of a doubtful law,5 but evenhandedness goes no further. It does not. guarantee success under Establishment Clause scrutiny.
Three cases permitting indirect aid to religion, Mueller v. Allen, 463 U. S. 388 (1983), Witters v. Washington Dept. of Servs. for Blind, 474 U. S. 481 (1986), and Zobrest v. Catalina Foothills School Dist., 509 U. S. 1 (1993), are among the latest of those to illustrate this relevance of evenhandedness when advancement is not so obvious as to be patently uncon*880stitutional. Each case involved a program in which benefits given to individuals on a religion-neutral basis ultimately were used by the individuals, in one way or another, to support religious institutions.6 In each, the fact that aid was distributed generally and on a neutral basis was a necessary condition for upholding the program at issue. Witters, supra, at 487-488; Mueller, supra, at 397-399; Zobrest, supra, at 10-11. But the significance of evenhandedness stopped there. We did not, in any of these cases, hold that satisfying the condition was sufficient, or dispositive. Even more importantly, we never held that evenhandedness might be sufficient to render direct aid to religion constitutional. Quite the contrary. Critical to our decisions in these cases was the fact that the aid was indirect; it reached religious institutions “only as a result of the genuinely independent and private choices of aid recipients,” Witters, supra, at 487; see also Mueller, supra, at 399-400; Zobrest, supra, at 10-13. In noting and relying on this particular feature of each of the programs at issue, we in fact reaffirmed the core prohibition on direct funding of religious activities. See Zobrest, supra, at 12-13; Witters, supra, at 487; see also Mueller, supra, at 399-400. Thus, our holdings in these cases were little more than extensions of the unremarkable proposition that “a State may issue a paycheck to one of its employees, who may then donate all or part of that paycheck to a religious institution, all without constitutional barrier . . . .” Witters, supra, at 486-487. Such “attenuated financial benefit[s], ultimately controlled by the private choices of indi*881vidual[s],” we have found, are simply not within the contemplation of the Establishment Clause’s broad prohibition. Mueller, supra, at 400; see also Witters, supra, at 493 (opinion of O’Connor, J.).7
*882Evenhandedness as one element of a permissibly attenuated benefit is, of course, a far cry from evenhandedness as a sufficient condition of constitutionality for direct financial support of religious proselytization, and our cases have unsurprisingly repudiated any such attempt to cut the Establishment Clause down to a mere prohibition against unequal direct aid. See, e. g., Tilton v. Richardson, 403 U. S., at 682-684 (striking portion of general aid program providing grants' for construction of college and university facilities to the extent program made possible the use of funds for sectarian activities);8 Wolman v. Walter, 433 U. S., at 252-255 (striking funding of field trips for nonpublic school students, such as are “provided to public school students in the district,” because of unacceptable danger that state funds would be used to foster religion). And nowhere has the Court’s adherence to the preeminence of the no-direct-funding principle over the principle of evenhandedness been as clear as in Bowen v. Kendrick, 487 U. S. 589 (1988).
Bowen involved consideration of the Adolescent Family Life Act (AFLA), a federal grant program providing funds to institutions for counseling and educational services related to adolescent sexuality and pregnancy. At the time of the litigation, 141 grants had been awarded under the AFLA to *883a broad array of both secular and religiously affiliated institutions. Id., at 597. In an Establishment Clause challenge to the Act brought by taxpayers and other interested parties, the District Court resolved the case on a pretrial motion for summary judgment, holding the AFLA program unconstitutional both on its face and also insofar as religious institutions were involved in receiving grants under the Act. When this Court reversed on the issue of facial constitutionality under the Establishment Clause, id., at 602-618, we said that there was “no intimation in the statute that at some point, or for some grantees, religious uses are permitted.” Id., at 614. On the contrary, after looking at the legislative history and applicable regulations, we found safeguards adequate to ensure that grants would not be “used by ... grantees in such a way as to advance religion.” Id., at 615.
With respect to the claim that the program was unconstitutional as applied, we remanded the ease to the District Court “for consideration of the evidence presented by appel-lees insofar as it sheds light on the manner in which the statute is presently being administered.” Id., at 621. Specifically, we told the District Court, on remand, to “consider . . . whether in particular cases AFLA aid has been used to fund ‘specifically religious activities] in an otherwise substantially secular setting.’ ” Ibid., quoting Hunt v. McNair, 413 U. S., at 743. In giving additional guidance to the District Court, we suggested that application of the Act would be unconstitutional if it turned out that aid recipients were using materials “that have an explicitly religious content or are designed to inculcate the views of a particular religious faith.” Bowen, 487 U. S., at 621. At no point in our opinion did we suggest that the breadth of potential recipients, or distribution on an evenhanded basis, could have justified the use of federal funds for religious activities, a position that would have made no sense after we had pegged the Act’s facial constitutionality to our conclusion that advancement of religion was not inevitable. Justice O’Connor’s separate *884opinion in the case underscored just this point: “I fully agree . . . that ‘[p]ublic funds may not be used to endorse the religious message.’ [487 U. S.,] at 642 [(Blackmun, J., dissenting)]. . . . [AJny use of public funds to promote religious doctrines violates the Establishment Clause.” Id., at 622-623 (concurring opinion) (emphasis in original).
Bowen was no sport; its pedigree was the line of Everson v. Board of Ed., 330 U. S., at 16-18, Board of Ed. v. Allen, 392 U. S., at 243-249, Tilton v. Richardson, supra, at 678-682, Hunt v. McNair, supra, at 742-745, and Roemer v. Board of Public Works of Md., 426 U. S., at 759-761. Each of these cases involved a general aid program that provided benefits to a broad array of secular and sectarian institutions on an evenhanded basis, but in none of them was that fact dispositive. The plurality opinion in Roemer made this point exactly:
“The Court has taken the view that a secular purpose and a facial neutrality may not be enough, if in fact the State is lending direct support to a religious activity. The State may not, for example, pay for what is actually a religious education, even though it purports to be paying for a secular one, and even though it makes its aid available to secular and religious institutions alike.” 426 U. S., at 747 (opinion of Blackmun, J.).
Instead, the central enquiry in each of these general aid cases, as in Bowen, was whether secular activities could be separated from the sectarian ones sufficiently to ensure that aid would flow to the secular alone.
Witters, Mueller, and Zobrest expressly preserve the standard thus exhibited so often. Each of these eases explicitly distinguished the indirect aid in issue from contrasting examples in the line of cases striking down direct aid, and each thereby expressly preserved the core constitutional principle that direct aid to religion is impermissible. See Zobrest, 509 U. S., at 11-13 (distinguishing Meek v. Pittenger, 421 U. S. 349 (1975), and School Dist. v. Ball, 473 U. S. 373 (1985), and noting that “ ‘[t]he State may not grant aid to a *885religious school, whether cash or in kind, where the effect of the aid is “that of a direct subsidy to the religious school” ’ ”) (quoting Witters, 474 U. S., at 487); see also ibid.; Mueller, 463 U. S., at 399. It appears that the University perfectly understood the primacy of the no-direct-funding rule over the evenhandedness principle when it drew the line short of funding “an[y] activity which primarily promotes or manifests a particular belief(s) in or about a deity or an ultimate reality.”9 App. to Pet. for Cert. 66a.
*886c
Since conformity with the marginal or limiting principle of evenhandedness is insufficient of itself to demonstrate the constitutionality of providing a government benefit that reaches religion, the Court must identify some further element in the funding scheme that does demonstrate its permissibility. For one reason or another, the Court’s chosen element appears to be the fact that under the University’s Guidelines, funds are sent to the printer chosen by Wide Awake, rather than to Wide Awake itself. Ante, at 842-844.
1
If the Court’s suggestion is that this feature of the funding program brings this case into line with Witters, Mueller, and Zobrest (discussed supra, at 879-881), the Court has misread those cases, which turned on the fact that the choice to benefit religion was made by a nonreligious third party standing between the government and a religious institution. See Witters, supra, at 487; see also Mueller, supra, at 399-400; Zobrest, supra, at 8-13. Here there is no third-party standing between the government and the ultimate religious beneficiary to break the circuit by its independent discretion to put state money to religious use. The printer, of course, has no option to take the money and use it to print a secular journal instead of Wide Awake. It only gets the money because of its contract to print a message of religious evangelism at the direction of Wide Awake, and it will receive payment only for doing precisely that. The formalism of distinguishing between payment to Wide Awake so it can pay an approved bill and payment of the approved bill itself cannot be the basis of a decision of constitutional law. If *887this indeed were a critical distinction, the Constitution would permit a State to pay all the bills of any religious institution;10 in fact, despite the Court’s purported adherence to the no-direct-funding principle, the State could simply hand out credit cards to religious institutions and honor the monthly statements (so long as someone could devise an evenhanded umbrella to cover the whole scheme). Witters and the other cases cannot be distinguished out of existence this way.
2
It is more probable, however, that the Court’s reference to the printer goes to a different attempt to justify the payment. On this purported justification, the payment to the printer is significant only as the last step in an argument resting on the assumption that a public university may give a religious group the use of any of its equipment or facilities so long as secular groups are likewise eligible. The Court starts with the cases of Widmar v. Vincent, 454 U. S. 263 (1981), Board of Ed. of Westside Community Schools (Dist. 66) v. Mergens, 496 U. S. 226 (1990), and Lamb’s Chapel v. Center Moriches Union Free School Dist., 508 U. S. 384 (1993), in which religious groups were held to be entitled to access for speaking in government buildings open generally for that purpose. The Court reasons that the availability of a forum has economic value (the government built and maintained the building, while the speakers saved the rent for a hall); and that economically there is no difference be*888tween the University’s provision of the value of the room and the value, say, of the University’s printing equipment; and that therefore the University must be able to provide the use of the latter. Since it may do that, the argument goes, it would be unduly formalistic to draw the line at paying for an outside printer, who simply does what the magazine’s publishers could have done with the University’s own printing equipment. Ante, at 843-844.
The argument is as unsound as it is simple, and the first of its troubles emerges from an examination of the cases relied upon to support it. The common factual thread running through Widmar, Mergens, and Lamb’s Chapel is that a governmental institution created a limited forum for the use of students in a school or college, or for the public at large, but sought to exclude speakers with religious messages. See generally Perry Ed. Assn. v. Perry Local Educators’ Assn., 460 U. S. 37, 45-46 (1983) (forum analysis). In each case the restriction was struck down either as an impermissible attempt to regulate the content of speech in an open forum (as in Widmar and Mergens) or to suppress a particular religious viewpoint (as in Lamb’s Chapel, see infra, at 897-898). In each case, to be sure, the religious speaker’s use of the room passed muster as an incident of a plan to facilitate speech generally for a secular purpose, entailing neither secular entanglement with religion nor risk that the religious speech would be taken to be the speech of the government or that the government’s endorsement of a religious message would be inferred. But each case drew ultimately on unexceptionable Speech Clause doctrine treating the evangelist, the Salvation Army, the millennialist, or the Hare Krishna like any other speaker in a public forum. It was the preservation of free speech on the model of the street corner that supplied the justification going beyond the requirement of evenhandedness.
The Court’s claim of support from these forum-access cases is ruled out by the very scope of their holdings. While *889they do indeed allow a limited benefit to religious speakers, they rest on the recognition that all speakers are entitled to use the street corner (even though the State paves the roads and provides police protection to everyone on the street) and on the analogy between the public street corner and open classroom space. Thus, the Court found it significant that the classroom speakers would engage in traditional speech activities in these forums, too, even though the rooms (like street corners) require some incidental state spending to maintain them. The analogy breaks down entirely, however, if the cases are read more broadly than the Court wrote them, to cover more than forums for literal speaking. There is no traditional street corner printing provided by the government on equal terms to all comers, and the forum cases cannot be lifted to a higher plane of generalization without admitting that new economic benefits are being extended directly to religion in clear violation of the principle barring direct aid. The argument from economic equivalence thus breaks down on recognizing that the direct state aid it would support is not mitigated by the street corner analogy in the service of free speech. Absent that, the rule against direct aid stands as a bar to printing services as well as printers.
3
It must, indeed, be a recognition of just this point that leads the Court to take a third tack, not in coming up with yet a third attempt at justification within the rules of existing case law, but in recasting the scope of the Establishment Clause in ways that make further affirmative justification unnecessary. Justice O’Connor makes a comprehensive analysis of the manner in which the activity fee is assessed and distributed. She concludes that the funding differs so sharply from religious funding out of governmental treasuries generally that it falls outside Establishment Clause’s purview in the absence of a message of religious endorsement (which she finds not to be present). Ante, at 849-852 (con*890curring opinion). The opinion of the Court concludes more expansively that the activity fee is not a tax, and then proceeds to find the aid permissible on the legal assumption that the bar against direct aid applies only to aid derived from tax revenue. I have already indicated why it is fanciful to treat the fee as anything but a tax, supra, at 873-874, and n. 3; see also ante, at 840 (noting mandatory nature of the fee), and will not repeat the point again. The novelty of the assumption that the direct aid bar only extends to aid derived from taxation, however, requires some response.
Although it was a taxation scheme that moved Madison to write in the first instance, the Court has never held that government resources obtained without taxation could be used for direct religious support, and our cases on direct government aid have frequently spoken in terms in no way limited to tax revenues. E. g., School Dist. v. Ball, 473 U. S., at 385 (“Although Establishment Clause jurisprudence is characterized by few absolutes, the Clause does absolutely prohibit government-financed or government-sponsored indoctrination into the beliefs of a particular religious faith”); Nyquist, 413 U. S., at 780 (“In the absence of an effective means of guaranteeing that the state aid derived from public funds will be used exclusively for secular, neutral, and non-ideological purposes, it is clear from our cases that direct aid, in whatever form is invalid”); id., at 772 (“Primary among' those evils” against which the Establishment Clause guards “have been sponsorship, financial support, and active involvement of the sovereign in religious activity”) (citations and internal quotation marks omitted); see also T. Curry, The First Freedoms 217 (1986) (At the time of the framing of the Bill of Rights, “[t]he belief that government assistance to religion, especially in the form of taxes, violated religious liberty had a long history”).
Allowing nontax funds to be spent on religion would, in fact, fly in the face of clear principle. Leaving entirely aside the question whether public nontax revenues could ever be used to finance religion without violating the endorsement *891test, see County of Allegheny v. American Civil Liberties Union, 492 U. S., at 593-594, any such use of them would ignore one of the dual objectives of the Establishment Clause, which was meant not only to protect individuals and their republics from the destructive consequences of mixing government and religion, but to protect religion from a corrupting dependence on support from the Government. Engel v. Vitale, 370 U. S. 421, 431 (1962) (the Establishment Clause’s “first and most immediate purpose rested on the belief that a union of government and religion tends to destroy government and to degrade religion”); Everson, 330 U. S., at 53 (Rutledge, J., dissenting) (“The great condition of religious liberty is that it be maintained free from sustenance, as also from other interferences, by the state. For when it comes to rest upon that secular foundation it vanishes with the resting”) (citing Madison’s Remonstrance ¶¶ 7, 8, reprinted in Everson, supra, at 63-72 (appendix to dissent of Rutledge, J.)); School Dist. of Abington Township v. Schempp, 374 U. S. 203, 259 (1963) (Brennan, J., concurring) (“It is not only the nonbeliever who fears the injection of sectarian doctrines and controversies into the civil polity, but in as high degree it is the devout believer who fears the secularization of a creed which becomes too deeply involved with and dependent upon the government”) (footnote omitted); Jefferson, A Bill for Establishing Religious Freedom, reprinted in 5 The Founder’s Constitution, at 84-85. Since the corrupting effect of government support does not turn on whether the Government’s own money comes from taxation or gift or the sale of public lands, the Establishment Clause could hardly relax its vigilance simply because tax revenue was not implicated. Accordingly, in the absence of a forthright disavowal, one can only assume that the Court does not mean to eliminate one half of the Establishment Clause’s justification.
D
Nothing in the Court’s opinion would lead me to end this enquiry into the application of the Establishment Clause any *892differently from the way I began it. The Court is ordering an instrumentality of the State to support religious evangelism with direct funding. This is a flat violation of the Establishment Clause.
II
Given the dispositive effect of the Establishment Clause’s bar to funding the magazine, there should be no need to decide whether in the absence of this bar the University would violate the Free Speech Clause by limiting funding as it has done. Widmar, 454 U. S., at 271 (university’s compliance with its Establishment Clause obligations can be a compelling interest justifying speech restriction). But the Court’s speech analysis may have independent application, and its flaws should not pass unremarked.
The Court acknowledges, ante, at 832, the necessity for a university to make judgments based on the content of what may be said or taught when it decides, in the absence of unlimited amounts of money or other resources, how to honor its educational responsibilities. Widmar, supra, at 276; cf. Perry, 460 U. S., at 49 (subject matter and speaker identity distinctions “are inherent and inescapable in the process of limiting a nonpublic forum to activities compatible with the intended purpose of the property”). Nor does the Court generally question that in allocating public funds a state university enjoys spacious discretion. Cf. Rust v. Sullivan, 500 U. S. 173, 194 (1991) (“[W]hen the government appropriates public funds to establish a program it is entitled to define the limits of that program”); Regan v. Taxation with Representation of Wash., 461 U. S. 540 (1983) (upholding government subsidization decision partial to one class of speaker).11 Ac*893cordingly, the Court recognizes that the relevant enquiry in this case is not merely whether the University bases its funding decisions on the subject matter of student speech; if there is an infirmity in the basis for the University’s funding decision, it must be that the University is impermissibly distinguishing among competing viewpoints, ante, at 829-830, citing, inter alia, Perry, supra, at 46; see also Lamb’s Chapel, 508 U. S., at 392-393 (subject-matter distinctions permissible in controlling access to limited public forum if reasonable and viewpoint neutral); Cornelius v. NAACP Legal Defense & Ed. Fund, Inc., 473 U. S. 788, 806 (1985) (similar); Regan, supra, at 548.12
The issue whether a distinction is based on viewpoint does not turn simply on whether a government regulation happens to be applied to a speaker who seeks to advance a particular viewpoint; the issue, of course, turns on whether the burden on speech is explained by reference to viewpoint. See Cornelius, supra, at 806 (“[T]he government violates the First Amendment when it denies access to a speaker solely *894to suppress the point of view he espouses on an otherwise includible subject”). As when deciding whether a speech restriction is content based or content neutral, “[t]he government’s purpose is the controlling consideration.” Ward v. Rock Against Racism, 491 U. S. 781, 791 (1989); see also ibid. (content neutrality turns on, inter alia, whether a speech restriction is “justified without reference to the content of the regulated speech”) (internal quotation marks and citations omitted) (emphasis deleted). So, for example, a city that enforces its excessive noise ordinance by pulling the plug on a rock band using a forbidden amplification system is not guilty of viewpoint discrimination simply because the band wishes to use that equipment to espouse antiracist views. Accord, Rock Against Racism, supra. Nor does a municipality’s decision to prohibit political advertising on bus placards amount to viewpoint discrimination when in the course of applying this policy it denies space to a person who wishes to speak in favor of a particular political candidate. Accord, Lehman v. Shaker Heights, 418 U. S. 298, 304 (1974) (plurality opinion).
Accordingly, the prohibition on viewpoint discrimination serves that important purpose of the Free Speech Clause, which is to bar the government from skewing public debate. Other things being equal, viewpoint discrimination occurs when government allows one message while prohibiting the messages of those who can reasonably be expected to respond. See First Nat. Bank of Boston v. Bellotti, 435 U. S. 765, 785-786 (1978) (“Especially where . . . the-legislature’s suppression of speech suggests an attempt to give one side of a debatable public question an advantage in expressing its views to the people, the First Amendment is plainly offended”) (footnote omitted); Madison Joint School Dist. No. 8 v. Wisconsin Employment Relations Comm’n, 429 U. S. 167, 175-176 (1976) (“To permit one side of a debatable public question to have a monopoly in expressing its views ... is the antithesis of constitutional guarantees”) (footnote omitted); *895United States v. Kokinda, 497 U. S. 720, 736 (1990) (viewpoint discrimination involves an “inten[t] to discourage one viewpoint and advance another”) (plurality opinion) (citations and internal quotation marks omitted). It is precisely this element of taking sides in a public debate that identifies viewpoint discrimination and makes it the most pernicious of all distinctions based on content. Thus, if government assists those espousing one point of view, neutrality requires it to assist those espousing opposing points of view, as well.
There is no viewpoint discrimination in the University’s application of its Guidelines to deny funding to Wide Awake. Under those Guidelines, a “religious activity],” which is not eligible for funding, App. to Pet. for Cert. 62a, is “an activity which primarily promotes or manifests a particular belief(s) in or about a deity or an ultimate reality,” id., at 66a. It is clear that this is the basis on which Wide Awake Productions was denied funding. Letter from Student Council to Ronald W. Rosenberger, App. 54 (“In reviewing the request by Wide Awake Productions, the Appropriations Committee determined your organization’s request could not be funded as it is a religious activity”). The discussion of Wide Awake’s content, supra, at 865-868, shows beyond any question that it “primarily promotes or manifests a particular belief(s) in or about a deity ... ,” in the very specific sense that its manifest function is to call students to repentance, to commitment to Jesus Christ, and to particular moral action because of its Christian character.
If the Guidelines were written or applied so as to limit only such Christian advocacy and no other evangelical efforts that might compete with it, the discrimination would be based on viewpoint. But that is not what the regulation authorizes; it applies to Muslim and Jewish and Buddhist advocacy as well as to Christian. And since it limits funding to activities promoting or manifesting a particular belief not only “in” but “about” a deity or ultimate reality, it applies to agnostics and atheists as well as it does to deists and theists *896(as the University maintained at oral argument, Tr. of Oral Arg. 18-19, and as the Court recognizes, see ante, at 836-837). The Guidelines, and their application to Wide Awake, thus do not skew debate by funding one position but not its competitors. As understood by their application to Wide Awake, they simply deny funding for hortatory speech that “primarily promotes or manifests” any view on the merits of religion; they deny funding for the entire subject matter of religious apologetics.
The Court, of course, reads the Guidelines differently, but while I believe the Court is wrong in construing their breadth, the important point is that even on the Court’s own construction the Guidelines impose no viewpoint discrimination. In attempting to demonstrate the potentially chilling effect such funding restrictions might have on learning in our Nation’s universities, the Court describes the Guidelines as “a sweeping restriction on student thought and student inquiry,” disentitling a vast array of topics to funding. Ante, at 836. As the Court reads the Guidelines to exclude “any writing that is explicable as resting upon a premise which presupposes the existence of a deity or ultimate reality,” ibid., as well as “those student journalistic efforts which primarily manifest or promote a belief that there is no deity and no ultimate reality,” the Court concludes that the major works of writers from Descartes to Sartre would be barred from the funding forum, ante, at 837. The Court goes so far as to suggest that the Guidelines, properly interpreted, tolerate nothing much more than essays on “making pasta or peanut butter cookies.” Ibid.
Now, the regulation is not so categorically broad as the Court protests. The Court reads the word “primarily” (“primarily promotes or manifests a particular belief(s) in or about a deity or an ultimate reality”) right out of the Guidelines, whereas it is obviously crucial in distinguishing between works characterized by the evangelism of Wide Awake and writing that merely happens to express views that a given religion might approve, or simply descriptive *897writing informing a reader about the position of a given religion. But, as I said, that is not the important point. Even if the Court were indeed correct about the funding restriction’s categorical breadth, the stringency of the restriction would most certainly not work any impermissible viewpoint discrimination under any prior understanding of that species of content discrimination. If a university wished to fund no speech beyond the subjects of pasta and cookie preparation, it surely would not be discriminating on the basis of someone’s viewpoint, at least absent some controversial claim that pasta and cookies did not exist. The upshot would be an instructional universe without higher education, but not a universe where one viewpoint was enriched above its competitors.
The Guidelines are thus substantially different from the access restriction considered in Lamb’s Chapel, the case upon which the Court heavily relies in finding a viewpoint distinction here, ante, at 830-832. Lamb’s Chapel addressed a school board’s regulation prohibiting the after-hours use of school premises “by any group for religious purposes,” even though the forum otherwise was open for a variety of social, civic, and recreational purposes. 508 U. S., at 387 (citation and internal quotation marks omitted). “Religious” was understood to refer to the viewpoint of a believer, and the regulation did not purport to deny access to any speaker wishing to express a nonreligious or expressly anti-religious point of view on any subject, see ibid. (“The issue in this case is whether ... it violates the Free Speech Clause of the First Amendment... to deny a church access to school premises to exhibit for public viewing and for assertedly religious purposes, a film series dealing with family and child-rearing issues”); id., at 394, citing May v. Evansville-Vanderburgh School Corp., 787 F. 2d 1105, 1114 (CA7 1986).13
*898With this understanding, it was unremarkable that in Lamb’s Chapel we unanimously determined that the access restriction, as applied to a speaker wishing to discuss family values from a Christian perspective, impermissibly distinguished between speakers on the basis of viewpoint. See Lamb’s Chapel, supra, at 393-394 (considering as-applied challenge only). Equally obvious is the distinction between that case and this one, where the regulation is being applied, not to deny funding for those who discuss issues in general from a religious viewpoint, but to those engaged in promoting or opposing religious conversion and religious observances as such. If this amounts to viewpoint discrimination, the Court has all but eviscerated the line between viewpoint and content.
To put the point another way, the Court’s decision equating a categorical exclusion of both sides of the religious debate with viewpoint discrimination suggests the Court has concluded that primarily religious and antireligious speech, grouped together, always provides an opposing (and not merely a related) viewpoint to any speech about any secular topic. Thus, the Court’s reasoning requires a university that funds private publications about any primarily nonreli*899gious topic also to fund publications primarily espousing adherence to or rejection of religion. But a university’s decision to fund a magazine about racism, and not to fund publications aimed at urging repentance before God does not skew the debate either about racism or the desirability of religious conversion. The Court’s contrary holding amounts to a significant reformulation of our viewpoint discrimination precedents and will significantly expand access to limited-access forums. See Greer v. Spock, 424 U. S. 828 (1976) (upholding regulation prohibiting political speeches on military base); Cornelius, 473 U. S., at 812 (exclusion from fundraising drive of political activity or advocacy groups is facially viewpoint neutral despite inclusion of charitable, health, and welfare agencies); Perry, 460 U. S., at 49-50, and n. 9 (ability of teachers’ bargaining representative to use internal school mail system does not require that access be provided to “any other citizen’s group or community organization with a message for school personnel”); Lehman, 418 U. S., at 304 (plurality opinion) (exclusion of political messages from forum permissible despite ability of nonpolitical speakers to use the forum).
►* — i H — H H-\
Since I cannot see the future I cannot tell, whether today’s decision portends much more than making a shambles out of student activity fees in public colleges. Still, my apprehension is whetted by Chief Justice Burger’s warning in Lemon v. Kurtzman, 403 U. S. 602, 624 (1971): “in constitutional adjudication some steps, which when taken were thought to approach ‘the verge,’ have become the platform for yet further steps. A certain momentum develops in constitutional theory and it can be a ‘downhill thrust’ easily set in motion but difficult to retard or stop.”
I respectfully dissent.
Justice Thomas suggests that Madison would have approved of the assessment bill if only it had satisfied the principle of evenhandedness. Nowhere in the Remonstrance, however, did Madison advance the view that Virginia, should be able to provide financial support for religion as part of a generally available subsidy program. Indeed, while Justice Thomas claims that the “funding provided by the Virginia assessment was to be extended only to Christian sects,” ante, at 855, it is clear that the bill was more general in scope than this. While the bill, which is reprinted in Everson v. Board of Ed. of Ewing, 330 U. S. 1, 72-74 (1947), provided that each taxpayer could designate a religious society to which he wanted his levy paid, id., at 73, it would also have allowed a taxpayer to refuse to appropriate his levy to any religious society, in which case the legislature was to use these unappropriated sums to fund “seminaries of learning.” Id., at 74 (contrary to Justice Thomas’s unsupported assertion, this portion of the bill was no less obligatory than any other). While some of these seminaries undoubtedly would have been religious in character, others would not- have been, as a seminary was generally understood at the time to be “any school, academy, college or university, in which young persons are instructed in the several branches of learning which may qualify them for their future employments.” N. Webster, An American Dictionary of the English Language (1st ed. 1828); see also 14 The Oxford English Dictionary 956 (2d ed. 1989). Not surprisingly, then, scholars have generally agreed that the bill would have provided funding for nonreligious schools. See, e. g., Laycock, “Nonpreferential” Aid to Religion: A False Claim About Original Intent, 27 Wm. & Mary L. Rev. 875, 897, and n. 108 (1986) (“Any taxpayer could refuse to designate a church, with un-designated church taxes going to a fund for schools. . . . The bill used the phrase ‘seminaries of learning,’ which almost certainly meant schools generally and not just schools for the training of ministers”); T. Buckley, *870Church and State in Revolutionary Virginia, 1776-1787, p. 133 (1977) (“The assessment had been carefully drafted to permit those who preferred to support education rather than religion to do so”); T. Curry, The First Freedoms 141 (1986) (“[T]hose taxes not designated for any specific denomination [were] allocated to education”). It is beside the point that “there was no system of public education in Virginia until several decades after the assessment bill was proposed,” ante, at 854, n. 1 (Thomas, J., concurring); because the bill was never passed, the funds that it would have made available for secular, public schools never materialized. The fact that the bill, if passed, would have funded secular as well as religious instruction did nothing to soften Madison’s opposition to it.
Nor is it fair to argue that Madison opposed the bill only because it treated religious groups unequally. Ante, at 854-855 (Thomas, J., concurring). In various paragraphs of the Remonstrance, Madison did complain about the bill’s peculiar burdens and exemptions, Everson, supra, at 66, but to identify this factor as the sole point of Madison’s opposition to the bill is unfaithful to the Remonstrance’s text. Madison strongly inveighed against the proposed aid for religion for a host of reasons (the Remonstrance numbers 15 paragraphs, each containing at least one point in opposition), and crucial here is the fact that many of those reasons would have applied whether or not the state aid was being distributed equally among sects, and whether or not the aid was going to those sects in the context of an evenhanded government program. See, e. g., Madison’s Remonstrance, reprinted in Everson, 330 U. S., at 64, ¶ 1 (“[I]n matters of Religion, no man’s right is abridged by the institution of Civil Society, and ... Religion is wholly exempt from its cognizance”); id., at 67, ¶ 6 (arguing that state support of religion “is a contradiction to the Christian Religion itself; for every page of it disavows a dependence on the powers of this world”); ibid., ¶ 7 (“[Experience witnesseth that ecclesiastical establishments, instead of maintaining the purity and efficacy of Religion, have had a contrary operation”). Madison’s objections were supplemented by numerous other petitions in opposition to the bill that likewise do not suggest that the lack of evenhandedness was its dispositive flaw. L. Levy, The Establishment Clause: Religion and the First Amendment 63-67 (2d ed. 1994). For example, the petition that received the largest number of signatories was motivated by the view that religion should only be supported voluntarily. Id., at 63-64. Indeed, Madison’s Remonstrance did not argue for a bill distributing aid to all sects and religions on an equal basis, and the *871outgrowth of the Remonstrance and the defeat of the Virginia assessment was not such a bill; rather, it was the Virginia Bill for Establishing Religious Freedom, which, as discussed in the text, proscribed the use of tax dollars for religious purposes.
In attempting to recast Madison’s opposition as having principally been targeted against “governmental preferences for particular religious faiths,” ante, at 856 (emphasis in original), Justice Thomas wishes to wage a battle that was lost long ago, for “this Court has rejected unequivocally the contention that the Establishment Clause forbids only governmental preference of one religion over another,” School Dist. of Abington Township v. Schempp, 374 U. S. 203, 216 (1963); see also Texas Monthly, Inc. v. Bullock, 489 U. S. 1, 17 (1989) (plurality opinion); id., at 28 (Blackmun, J., concurring in judgment); Wallace v. Jaffree, 472 U. S. 38, 52-53 (1985); Torcaso v. Watkins, 367 U. S. 488, 495 (1961); Engel v. Vitale, 370 U. S. 421, 430 (1962); Everson, supra, at 15; see generally Lee v. Weisman, 505 U. S. 577, 609-616 (1992) (Souter, J., concurring).
Justice Thomas attempts to cast doubt on this accepted version of Establishment Clause history by reference to historical facts that are largely inapposite. Ante, at 857-858, 862-863 (concurring opinion). As I have said elsewhere, individual Acts of Congress, especially when they are few and far between, scarcely serve as an authoritative guide to the meaning of the Religion Clauses, for “like other politicians, [members of the early Congresses] could raise constitutional ideals one day and turn their backs on them the next. [For example,] . . . [t]en years after proposing the First Amendment, Congress passed the Alien and Sedition Acts, measures patently unconstitutional by modern standards. If the early Congress’s political actions were determinative, and not merely relevant, evidence of constitutional meaning, we would have to gut our current First Amendment doctrine to make room for political censorship.” Lee v. Weisman, supra, at 626 (concurring opinion). The legislation cited by Justice Thomas, including the Northwest Ordinance, is no more dispositive than the Alien and Sedition Acts in interpreting the First Amendment. Even less persuasive, then, are citations to constitutionally untested Acts dating from the mid-19th century, for without some rather innovative argument, they cannot be offered as providing an authoritative gloss on the Framers’ intent.
Justice Thomas’s references to Madison’s actions as a legislator provide little support for his cause. Justice Thomas seeks to draw a significant lesson out of the fact that, in seeking to disestablish the Anglican Church in Virginia in 1776, Madison did not inveigh against state funding of religious activities. Ante, at 857 (concurring opinion). That was *873not the task at hand, however. Madison was acting with the specific goal of eliminating the special privileges enjoyed by Virginia Anglicans, and he made no effort to lay out the broader views of church and state that came to bear in his drafting of the First Amendment some 13 years later. That Madison did not speak in more expansive terms than necessary in 1776 was hardly surprising for, as it was, his proposal was defeated by the Virginia Convention as having gone too far. Ibid.
Similarly, the invocation of Madison’s tenure on the congressional committee that approved funding for legislative chaplains provides no support for more general principles that run counter to settled Establishment Clause jurisprudence. As I have previously pointed out, Madison, upon retirement, “insisted that ‘it was not with my approbation, that the deviation from [the immunity of religion from civil jurisdiction] took place in Congs., when they appointed Chaplains, to be paid from the Natl. Treasury.’ ” Lee, 505 U. S., at 625, n. 6, quoting Letter from J. Madison to E. Livingston (July 10, 1822), in 5 The Founders’ Constitution 105 (P. Kurland & R. Lerner eds. (1987)). And when we turned our attention to deciding whether funding of legislative chaplains posed an establishment problem, we did not address the practice as one instance of a larger class of permissible government funding of religious activities. Instead, Marsh v. Chambers, 463 U. S. 783, 791 (1983), explicitly relied on the singular, 200-year pedigree of legislative chaplains, noting that “[t]his unique history” justified carving out an exception for the specific practice in question. Given that the decision upholding this practice was expressly limited to its facts, then, it would stand the Establishment Clause on its head to extract from it a broad rule permitting the funding of religious activities.
In the District Court, the parties agreed to the following facts: “The University of Virginia has charged at all times relevant herein and currently charges each full-time student a compulsory student activity fee of $14.00 per semester. There is no procedural or other mechanism by which a student may decline to pay the fee.” App. 37; see also id., at 9, 21.
To the extent the Court perceives some distinction between the printing and dissemination of evangelism and proselytization, and core religious activity “in [its] usual sense,” ante, at 844, this distinction goes entirely unexplained in the Court’s opinion.
In a narrow band of cases at the polar extreme from direct funding cases, those involving essential public benefits commonly associated with living in an organized society (like police and fire protection, for example), evenhandedness may become important to ensuring that religious interests are not inhibited.
In Zobrest, a deaf student sought to have an interpreter, provided under a state Act aiding individuals with disabilities, accompany him to a Roman Catholic high school. In Witters, a blind student sought to use aid, provided under a state program for assistance to handicapped persons, to attend a private Christian college. In Mueller, parents sought to take a tax deduction, available for parents of both public and nonpublic schoolchildren, for certain expenses incurred in connection with providing education for their children in private religious schools. *882on Bittker ... is misplaced in this context,” ante, at 860, n. 5, is not on point. Even granting that Justice Thomas’s assertion of equivalence is reasonable, he cannot and does not deny the fact that the Court in Walz explicitly distinguished tax exemptions from direct money subsidies, 397 U. S., at 675, and rested its decision on that distinction. If Justice Thomas’s assertion of equivalence should prevail then the Walz Court necessarily was wrong about a distinction critical to its holding. Justice Thomas can hardly use Walz coherently for support after removing the basis on which it relies.
Walz v. Tax Comm’n of City of New York, 397 U. S. 664 (1970), is yet another example of a case in which the Court treated the general availability of a government benefit as a significant condition defining compliance with the Establishment Clause, but did not deem that condition sufficient. In upholding state property tax exemptions given to religious organizations in Walz, we noted that the law at issue was applicable to “a broad class of property owned by nonprofit [and] quasi-public corporations,” id., at 673, but did not rest on that factor alone. Critical to our decision was the central principle that direct funding of religious activities is prohibited under the Establishment Clause. “It is sufficient to note that for the men who wrote the Religion Clauses of the First Amendment the ‘establishment’ of a religion connoted sponsorship, financial support, and active involvement of the sovereign in religious activity.” Id., at 668. We emphasized that the tax exemptions did not involve the expenditure of government funds in support of religious activities. “The grant of a tax exemption is not sponsorship since the government does not transfer part of its revenue to churches but simply abstains from demanding that the church support the state.” Id., at 675. Moreover, we noted that in the property taxation context, “exemption[s] creat[e] only a minimal and remote involvement between church and. state and. far less- than taxation of churches,” and in operation "ten[d] to complement and reinforce the desired separation insulating” church and state, idl, at 676;: and that religious property tax exemptions have been in place for over 200. years without disruption to the interests represented by the Establishment Clause, id., at 676-680.
Justice Thomas’s assertion, that “[a] tax exemption in many cases is economically and functionally indistinguishable from a direct monetary subsidy,” ante, at 859 (concurring opinion) (footnote omitted), assumes that the “natural” or “correct” tax base is so self-evident that any provision excusing a person or institution from taxes to which others are subjected must be a departure from the natural tax base rather than part of the definition of the tax base itself. The equivalence (asserted by Justice Thomas, ibid.) between a direct money subsidy and the tax liability avoided by an institution (because it is part of the class of institutions that defines the relevant tax base by its exclusion) was tested and dispatched long ago by Professor Bittker in Churches, Taxes and the Constitution, 78 Yale L. J. 1285 (1969). Justice Thomas’s suggestion that my “reliance
Although the main opinion in Tilton was a plurality, the entire Court was unanimous on this point. See 403 U. S., at 682-684 (plurality opinion); id., at 692 (Douglas, J., joined by Black and Marshall, JJ., concurring in part and dissenting in part); Lemon v. Kurtzman, 403 U. S. 602, 659-661 (1971) (opinion of Brennan, J.); id., at 665, n. 1 (opinion of White, J.).
Congress apparently also reads our eases as the University did, for it routinely excludes religious activities from general funding programs. See, e. g., 20 U. S. C. § 1062(b) (federal grant program for institutions of higher education; “[n]o grant may be made under this chapter for any educational program, activity, or service related to sectarian instruction or religious worship, or provided by a school or department of divinity”); 20 U. S. C. § 1069c (certain grants to higher education institutions “may not be used ... for a school or department of divinity or any religious worship or sectarian activity ...”); 20 U. S. C. § 1132c-3(c) (1988 ed., Supp. V) (federal assistance for renovation of certain academic facilities; “[n]o loan may be made under this part for any educational program, activity or service related to sectarian instruction or religious worship or provided by a school or department of divinity or to an institution in which a substantial portion of its functions is subsumed in a religious mission”); 20 U. S. C. § 1132i(c) (grant program for educational facilities; “no project assisted with funds under this subchapter shall ever be used for religious worship or a sectarian activity or for a school or department of divinity”); 20 U. S. C. § 1213d (“No grant may be made under this chapter for any educational program, activity, or service related to sectarian instruction or religious worship, or provided by a school or department of divinity”); 25 U. S. C. § 3306(a) (1988 ed., Supp. V) (funding for Indian higher education programs; “[n]one of the funds made available under this subchapter may be used for study at any school or department of divinity or for any religious worship or sectarian activity”); 29 U. S. C. § 776(g) (grants for projects and activities for rehabilitation of handicapped persons; “[n]o funds provided under this subchapter may be used to assist in the construction of any facility which is or will be used for religious worship or any sectarian activity”); 42 U. S. C. § 3027(a)(14)(A)(iv) (1988 ed. and Supp. V) (requiring States seeking federal aid for construction of centers for the elderly to submit plans providing assurances that “the facilities] will not be used and [are] not intended to be used for sectarian instruction or as ... place[s] for religious worship”); 42 U. S. C. § 5001(a)(2) (1988 ed., *886Supp. V) (federal grants to support volunteer projects for the elderly, but not including “projects involving the construction, operation, or maintenance of so much of any facility used or to be used for sectarian instruction or as a place for religious worship”); 42 U. S. C. § 9858k(a) (1988 ed., Supp. V) (no child care and development block grants “shall be expended for any sectarian purpose or activity, including sectarian worship or instruction”).
The Court acknowledges that “if the State pays a church’s bills it is subsidizing it,” and concedes that “we must guard against this abuse.” Ante, at 844. These concerns are not present here, the Court contends, because Wide Awake “is not a religious institution, at least in the usual sense of that term as used in our ease law.” Ibid. The Court’s concession suggests that its distinction between paying a religious institution and paying a religious institution’s bills is not really significant. But if the Court is relying on its characterization of Wide Awake as not a religious institution, “at least in the usual sense,” the Court could presumably stop right there.
The Court draws a distinction between a State’s use of public funds to advance its own speech and the State’s funding of private speech, suggesting that authority to make content-related choices is at its most powerful when the State undertakes the former. Ante, at 833-835. I would not argue otherwise, see Hazelwood School Dist. v. Kuhlmeier, 484 U. S. 260, 270-273 (1988), but I do suggest that this case reveals the difficulties that can be encountered in drawing this distinction. There is a communi*893cative element inherent in the very act of funding itself, cf. Buckley v. Valeo, 424 U. S. 1, 15-19 (1976) (per curiam), and although it is the student speakers who choose which particular messages to advance in the forum created by the University, the initial act of defining the boundaries of the forum is a decision attributable to the University, not the students. In any event, even assuming that private and state speech always may be separated by clean lines and that this case involves only the former, I believe the distinction is irrelevant here because, as is discussed infra, this case does not involve viewpoint discrimination.
1 do not decide that all viewpoint discrimination in a public university’s funding determinations would violate the Free Speech Clause. If, however, the determinations are made on the basis of a reasonable subject-matter distinction, but not on a viewpoint distinction, there is no violation. In a limited-access forum, a speech restriction must be “ ‘reasonable in light of the purpose served by the forum’ ” as well as viewpoint neutral. E. g., Lamb’s Chapel, 508 U. S., at 392-393, quoting Cornelius, 473 U. S., at 806. Because petitioners have not challenged the University’s Guideline as unreasonable, I express no opinion on that or on the question whether the reasonableness criterion applies in speech funding eases in the same manner that it applies in limited-access forum cases.
See also Tr. of Oral Arg. in Lamb’s Chapel v. Center Moriches Union Free School Dist., O. T. 1992, No. 91-2024, where counsel for the school district charged with enforcing the restriction unequivocally admitted that anyone with an atheistic or antireligious message would be permitted to *898use school property under the rules of the forum. Id., at 47, 57-68. The complete exchange during the oral argument in Lamb’s Chapel went as follows:
“QUESTION: But do I understand your statement you made earlier that supposing you had a communist group that wanted to address the subject of family values and they thought there was a value in not having children waste their time going to Sunday school or church and therefore they had a point of view that was definitely antireligious, they would be permitted, under your policy, to discuss family values in that context?
“[COUNSEL]: Yes. Yes, Your Honor, that’s correct.
“QUESTION: Counsel, in your earlier discussions with [the Court] you indicated that communists would be able to give their perspective on family. I — I assume from that that atheists would be able to give theirs under your rules.
“[COUNSEL]: Yes, Your Honor.”