with whom Justice O’Connor joins, concurring in the judgment.

The Texas statute at issue touches upon values that underlie three different Clauses of the First Amendment: the Free Exercise Clause, the Establishment Clause, and the Press Clause. As indicated by the number of opinions issued in this case today, harmonizing these several values is not an easy task.

The Free Exercise Clause value suggests that a State may not impose a tax on spreading the gospel. See Follett v. McCormick, 321 U. S. 573 (1944), and Murdock v. Pennsylvania, 319 U. S. 105 (1943). The Establishment Clause value suggests that a State may not give a tax break to those who spread the gospel that it does not also give to others who actively might advocate disbelief in religion. See Torcaso v. Watkins, 367 U. S. 488, 495 (1961); Everson v. Board of Education of Ewing, 330 U. S. 1, 15-16 (1947). The Press Clause value suggests that a State may not tax the sale of some publications, but not others, based on their content, absent a compelling reason for doing so. See Arkansas Writers’ Project, Inc. v. Ragland, 481 U. S. 221, 231 (1987).

It perhaps is fairly easy to reconcile the Free Exercise and Press Clause values. If the Free Exercise Clause suggests that a State may not tax the sale of religious literature by a religious organization, this fact alone would give a State a compelling reason to exclude this category of sales from an otherwise general sales tax. In this respect, I agree gener*27ally with what Justice Sc alia says in Part II of his dissenting opinion.

I find it more difficult to reconcile in this case the Free Exercise and Establishment Clause values. The Free Exercise Clause suggests that a special exemption for religious books is required. The Establishment Clause suggests that a special exemption for religious books is forbidden. This tension between mandated and prohibited religious exemptions is well recognized. See, e. g., Walz v. Tax Comm’n of New York City, 397 U. S. 664, 668-669 (1970). Of course, identifying the problem does not resolve it.

Justice Brennan’s opinion, in its Part IV, would resolve the tension between the Free Exercise and Establishment Clause values simply by subordinating the Free Exercise value, even, it seems to me, at the expense of longstanding, precedents. See ante, at 21-25 (repudiating Follett and Murdock to the extent inconsistent with the newfound proposition that a State generally may tax the sale of a Bible by a church). Justice Scalia’s opinion, conversely, would'subordinate the Establishment Clause value. This position, it seems to me, runs afoul of the previously settled notion that government may not favor religious belief over disbelief. See, e. g., Wallace v. Jaffree, 472 U. S. 38, 53 (1985); Welsh v. United States, 398 U. S. 333, 356 (1970) (Harlan, J., concurring in result); Epperson v. Arkansas, 393 U. S. 97, 103-104 (1968); Abington School District v. Schempp, 374 U. S. 203, 218, 220 (1963); Torcaso v. Watkins, 367 U. S., at 495.

Perhaps it is a vain desire, but I would like to decide the present case without necessarily sacrificing either the Free Exercise Clause value or the Establishment Clause value. It is possible for a State to write a tax-exemption statute consistent with both values: for example, a state statute might exempt the sale not only of religious literature distributed by a religious organization but also of philosophical literature distributed by nonreligious organizations devoted to such matters of conscience as life and death, good and evil, being *28and nonbeing, right and wrong. Such a statute, moreover, should survive Press Clause scrutiny because its exemption would be narrowly tailored to meet the compelling interests that underlie both the Free Exercise and Establishment Clauses.

To recognize this possible reconciliation of the competing First Amendment considerations is one thing; to impose it upon a State as its only legislative choice is something else. Justice Scalia rightly points out, post, at 42, that the Free Exercise and Establishment Clauses often appear like Scylla and Charybdis, leaving a State little room to maneuver between them. The Press Clause adds yet a third hazard to a State’s safe passage through the legislative waters concerning the taxation of books and journals. We in the Judiciary must be wary of interpreting these three constitutional Clauses in a manner that negates the legislative role altogether.

I believe we can avoid most of these difficulties with a narrow resolution of the case before us. We need not decide today the extent to which the Free Exercise Clause requires a tax exemption for the sale of religious literature by a religious organization; in other words, defining the ultimate scope of Follett and Murdock may be left for another day. We need decide here only whether a tax exemption limited to the sale of religious literature by religious organizations violates the Establishment Clause. I conclude that it does.

In this case, by confining the tax exemption exclusively to the sale of religious publications, Texas engaged in preferential support for the communication of religious messages. Although some forms of accommodating religion are constitutionally permissible, see Corporation of Presiding Bishop of Church of Jesus Christ of Latter-day Saints v. Amos, 483 U. S. 327 (1987), this one surely is not. A statutory preference for the dissemination of religious ideas offends our most basic understanding of what the Establishment Clause is all about and hence is constitutionally intolerable. See Wallace *29v. Jaffree, 472 U. S., at 69-70 (O’Connor, J., concurring in judgment); Epperson v. Arkansas, 393 U. S., at 103-104. Accordingly, whether or not Follett and Murdock prohibit taxing the sale of religious literature, the Establishment Clause prohibits a tax exemption limited to the sale of religious literature. Cf. Estate of Thornton v. Caldor, Inc., 472 U. S. 703 (1985) (the Establishment Clause prohibits a statute that grants employees an unqualified right not to work on their Sabbath), and Hobbie v. Unemployment Appeals Comm’n of Fla., 480 U. S. 136, 145-146, and n. 11 (1987) (consistent-with Caldor, the Free Exercise Clause prohibits denying unemployment compensation to employees who refuse to work on their Sabbath).

At oral argument, appellees suggested that the statute at issue here exempted from taxation the sale of atheistic literature distributed by an atheistic organization. Tr. of Oral Arg. 33. If true, this statute might survive Establishment Clause scrutiny, as well as Free Exercise and Press Clause scrutiny. But, as appellees were quick to concede at argument, the record contains nothing to support this facially implausible interpretation of the statute. Ibid. Thus, constrained to construe this Texas statute as exempting religious literature alone, I concur in the holding that it contravenes the Establishment Clause, and in remanding the case for further proceedings not inconsistent with this holding.

announced the judgment of the Court and delivered an opinion, in which Justice Marshall and Justice Stevens join.

Texas exempts from its sales tax “[pjeriodicals that are published or distributed by a religious faith and that consist wholly of writings promulgating the teaching of the faith and books that consist wholly of writings sacred to a religious faith.” Tex. Tax Code Ann. §151.312 (1982). The question presented is whether this exemption violates the Establishment Clause or the Free Press Clause of the First Amendment when the State denies a like exemption for other publications. We hold that, when confined exclusively to publications advancing the tenets of a religious faith, the exemption runs afoul of the Establishment Clause; accordingly, we need not reach the question whether it contravenes the Free Press Clause as well.

I

Prior to October 2, 1984, Texas exempted from its sales and use tax magazine subscriptions running half a year or longer and entered as second class mail. Tex. Tax Code Ann. §151.320 (1982). This exemption was repealed as of October 2, 1984, before being reinstated effective October 1, 1987. Tex. Tax Code Ann. §151.320 (Supp. 1988-1989). Throughout this 3-year period, Texas continued to exempt from its sales and use tax periodicals published or distributed by a religious faith consisting entirely of writings promulgating the teaching of the faith, along with books consisting *6solely of writings sacred to a religious faith. Tex. Tax Code Ann. § 151.312 (1982).

Appellant Texas Monthly, Inc., publishes a general interest magazine of the same name. Appellant is not a religious faith, and its magazine does not contain only articles promulgating the teaching of a religious faith. Thus, it was required during this 3-year period to collect and remit to the State the applicable sales tax on the price of qualifying subscription sales. Tex. Tax Code Ann. §§ 151.051, 151.052, 151.401 (1982 and Supp. 1988-1989). In 1985, appellant paid sales taxes of $149,107.74 under protest and sued to recover those payments in state court.

The District Court of Travis County, Texas, ruled that an exclusive exemption for religious periodicals had “no basis . . . other than the promotion of religion itself, a prohibited reason” under the Establishment Clause. App. to Juris. Statement 47. The court also found the exemption unconstitutional because it discriminated on the basis of the content of publications, presumably in violation of the Free Press Clause. Id., at 42. Declaring itself “without power to rewrite the statute to make religious periodicals subject to tax,” id., at 47, the court struck down the tax as applied to nonreligious periodicals and ordered the State to refund the amount of tax Texas Monthly had paid, plus interest. Id., at 43.

The Court of Appeals, Third Supreme Judicial District of Texas, reversed by a 2-to-l vote. 731 S. W. 2d 160 (1987). Applying the tripartite test enunciated in Lemon v. Kurtzman, 403 U. S. 602, 612-613 (1971), the court held, first, that the exemption served the secular purpose of preserving separation between church and state. Second, the court asserted that the exemption did not have the primary effect of advancing or inhibiting religion, because “the effect of religious tax exemptions such as § 151.312 is to permit religious organizations to be independent of government support or sanction.” 731 S. W. 2d, at 163. The court considered it irrele*7vant that the exemption did not extend to other nonprofit or secular publications, because “the neutrality toward religion effected by the grant of an exemption for religious periodicals” remained unaffected by the provision or denial of a similar exemption for nonreligious publications. Id., at 164. Finally, the court concluded that the exemption did not produce impermissible government entanglement with religion. Rather than scrutinize each publication for which a publisher sought an exemption for conformity with the statute’s terms, the court found, the Comptroller’s Office merely required that a group applying for an exemption demonstrate that it was a religious organization. Once a satisfactory showing had been made, the Comptroller’s Office did not later reassess the group’s status as a religious organization. It further allowed the group to determine, without review by the State, which of its publications promulgated the teaching of its faith. Because the exemption was administered to minimize state entanglement with religion, the court thought it consistent with Lemon’s third prong.

In addition, the court rejected Texas Monthly’s claim that the exemption violated the Free Press Clause because it discriminated among publications on the basis of their content. The court read our decision in Arkansas Writers’ Project, Inc. v. Ragland, 481 U. S. 221 (1987), to preclude only those taxes that are imposed solely on the press or targeted at a small group within the press. Because Texas’ exemption encompassed only a minority of publications, leaving the bulk of subscription sales subject to tax, the court reasoned that it escaped the strictures of the Free Press Clause as we had interpreted it.

now reverse. We noted probable jurisdiction, 485 U. S. 958 (1988), and now reverse.

As a preliminary matter, Texas argues that appellant lacks standing to challenge the constitutionality of the exemption. It claims that if this Court were to declare the exemption *8invalid, the proper course under state law would be to remove the exemption for religious publications, rather than extend it to nonreligious periodicals or strike down the sales and use tax in its entirety. If Texas is right, appellant cannot obtain a refund of the tax it paid under protest. Nor can it qualify for injunctive relief, because.its subscription sales are no longer taxed. Hence, Texas contends, appellant cannot show that it has suffered or is threatened with redressable injury, which this Court declared to be a prerequisite for standing in Valley Forge Christian College v. Americans United for Separation of Church & State, Inc., 454 U. S. 464, 472 (1982).

The State’s contention is misguided. In Arkansas Writers’ Project, supra, at 227, we rejected a similar argument, “for it would effectively insulate underinclusive statutes from constitutional challenge, a proposition we soundly rejected in Orr v. Orr, 440 U. S. 268, 272 (1979).” It is not for us to decide whether the correct response as a matter of state law to a finding that a state tax exemption is unconstitutional is to eliminate the exemption, to curtail it, to broaden it, or to invalidate the tax altogether. Nor does it make any difference — contrary to the State’s suggestion — that Texas Monthly seeks only a refund and not prospective relief, as did the appellant in Arkansas Writers’ Project. A live controversy persists over Texas Monthly’s right to recover the $149,107.74 it paid, plus interest. Texas cannot strip appellant of standing by changing the law after taking its money.

Ill

In proscribing all laws “respecting an establishment of religion,” the Constitution prohibits, at the very least, legislation that constitutes an endorsement of one or another set of religious beliefs or of religion generally. It is part of our settled jurisprudence that “the Establishment Clause prohibits government from abandoning secular purposes in order to put an imprimatur on one religion, or on religion as such, or *9to favor the adherents of any sect or religious organization.” Gillette v. United States, 401 U. S. 437, 450 (1971). See, e. g., School Dist. of Grand Rapids v. Ball, 473 U. S. 373, 381 (1985); Wallace v. Jaffree, 472 U. S. 38, 52-53, and n. 37 (1985); Welsh v. United States, 398 U. S. 333, 356-357 (1970) (Harlan, J., concurring in result); Epperson v. Arkansas, 393 U. S. 97, 103-104 (1968); Abington School Dist. v. Schempp, 374 U. S. 203, 216-217 (1963); Torcaso v. Watkins, 367 U. S. 488, 495 (1961); Everson v. Board of Education of Ewing, 330 U. S. 1, 15-16 (1947). The core notion animating the requirement that a statute possess “a secular legislative purpose” and that “its principal or primary effect... be one that neither advances nor inhibits religion,” Lemon v. Kurtzman, 403 U. S., at 612, is not only that government may not be overtly hostile to religion but also that it may not place its prestige, coercive authority, or resources behind a single religious faith or behind religious belief in general, compelling nonadherents to support the practices or proselytizing of favored religious organizations and conveying the message that those who do not contribute gladly are less than full members of the community.1

*10It does not follow, of course, that government policies with secular objectives may not incidentally benefit religion. The nonsectarian aims of government and the interests of religious groups often overlap, and this Court has never required that public authorities refrain from implementing reasonable measures to advance legitimate secular goals merely because they would thereby relieve religious groups of costs they would otherwise incur. See Mueller v. Allen, 463 U. S. 388, 393 (1983). Nor have we required that legislative categories make no explicit reference to religion. See Wallace v. Jaffree, supra, at 70 (O’Connor, J., concurring in judgment) (“The endorsement test does not preclude government from acknowledging religion or from taking religion into account in making law and policy”); Lynch v. Donnelly, 465 U. S. 668, 715 (1984) (Brennan, J., dissenting). Government need not resign itself to ineffectual diffidence because of exaggerated fears of contagion of or by religion, so long as neither intrudes unduly into the affairs of the other.

Thus, in Widmar v. Vincent, 454 U. S. 263 (1981), we held that a state university that makes its facilities available to registered student groups may not deny equal access to a registered student group desiring to use those facilities for religious worship or discussion. Although religious groups benefit from access to university facilities, a state university may not discriminate against them based on the content of their speech, and the university need not ban all student group meetings on campus in order to avoid providing any assistance to religion. Similarly, in Mueller v. Allen, supra, we upheld a state income tax deduction for the cost of tuition, transportation, and nonreligious textbooks paid by a taxpayer for the benefit of a dependent. To be sure, the deduction aided parochial schools and parents whose children attended them, as well as nonsectarian private schools and their pupils’ parents. We did not conclude, however, that *11this subsidy deprived the law of an overriding secular purpose or effect. And in the case most nearly on point, Walz v. Tax Comm’n of New York City, 397 U. S. 664 (1970), we sustained a property tax exemption that applied to religious properties no less than to real estate owned by a wide array of nonprofit organizations, despite the sizable tax savings it accorded religious groups.

In all of these cases, however, we emphasized that the benefits derived by religious organizations flowed to a large number of nonreligious groups as well. Indeed, were those benefits confined to religious organizations, they could not have appeared other than as state sponsorship of religion; if that were so, we would not have hesitated to strike them down for lacking a secular purpose and effect. See, e. g., School Dist. of Grand Rapids v. Ball, supra (invalidating state-funded educational programs in private schools, where 40 of the 41 beneficiaries were religious schools); Estate of Thornton v. Caldor, Inc., 472 U. S. 703 (1985) (finding vio-lative of the Establishment Clause a statute providing Sabbath observers with an unconditional right not to work on their chosen Sabbath).

In Widmar v. Vincent, we noted that an open forum in a public university would not betray state approval of religion so long as the forum was available “to a broad class of nonreligious as well as religious speakers.” 454 U. S., at 274. “The provision of benefits to so broad a spectrum of groups,” we said, “is an important index of secular effect.” Ibid. We concluded that the primary effect of an open forum would not be to advance religion, “[a]t least in the absence of empirical evidence that religious groups will dominate” it. Id., at 275. Likewise, in Mueller v. Allen, we deemed it “particularly significant,” 463 U. S., at 396, that “the deduction is available for educational expenses incurred by all parents, including those whose children attend public schools and those whose children attend nonsectarian private schools or sectarian private schools.” Id., at 397.

*12Finally, we emphasized in Walz that in granting a property tax deduction, the State “has not singled out one particular church or religious group or even churches as such; rather, it has granted exemption to all houses of religious worship within a broad class of property owned by nonprofit, quasi-public corporations which include hospitals, libraries, playgrounds, scientific, professional, historical, and patriotic groups.” 397 U. S., at 673. The breadth of New York’s property tax exemption was essential to our holding that it was “not aimed at establishing, sponsoring, or supporting religion,” id., at 674, but rather possessed the legitimate secular purpose and effect of contributing to the community’s moral and intellectual diversity and encouraging private groups to undertake projects that advanced the community’s well-being and that would otherwise have to be funded by tax revenues or left undone.2 Moreover, “[t]he scheme [was] *13not designed to inject any religious activity into a nonreligious context, as was the case with school prayers. No particular activity of a religious organization — for example, the propagation of its beliefs — [was] specially promoted by the exemptions.” Id., at 689 (Brennan, J., concurring). As Justice Harlan observed:

“To the extent that religious institutions sponsor the secular activities that this legislation is designed to promote, it is consistent with neutrality to grant them an exemption just as other organizations devoting resources to these projects receive exemptions. ... As long as the breadth of exemption includes groups that pursue cultural, moral, or spiritual improvement in multifarious secular ways, including, I would suppose, groups whose avowed tenets may be antitheological, atheistic, or agnostic, I can see no lack of neutrality in extending the benefit of the exemption to organized religious groups.”3 Id., at 697 (separate opinion) (footnote omitted).

*14Texas’ sales tax exemption for periodicals published or distributed by a religious faith and consisting wholly of writings promulgating the teaching of the faith lacks sufficient breadth to pass scrutiny under the Establishment Clause. Every tax exemption constitutes a subsidy that affects non-qualifying taxpayers, forcing them to become “indirect and vicarious ‘donors.’” Bob Jones University v. United States, 461 U. S. 574, 591 (1983). See also Regan v. Taxation with Representation of Wash., 461 U. S. 540, 544 (1983). Insofar as that subsidy is conferred upon a wide array of nonsectarian groups as well as religious organizations in pursuit of some legitimate secular end,4 the fact that religious groups *15benefit incidentally does not deprive the subsidy of the secular purpose and primary effect mandated by the Establishment Clause. However, when government directs a subsidy exclusively to religious organizations that is not required by the Free Exercise Clause and that either burdens nonbenefi-ciaries markedly or cannot reasonably be seen as removing a significant state-imposed deterrent to the free exercise of religion, as Texas has done, see infra, at 17-20, it “provide[s] unjustifiable awards of assistance to religious organizations” and cannot but “conve[y] a message of endorsement” to slighted members of the community. Corporation of Presiding Bishop of Church of Jesus Christ of Latter-day Saints v. Amos, 483 U. S. 327, 348 (1987) (O’Connor, J., concurring in judgment). This is particularly true where, as here, the subsidy is targeted at writings that promulgate the teachings of religious faiths.5 It is difficult to view Texas’ narrow exemption as anything but state sponsorship of religious belief, regardless of whether one adopts the perspective of beneficiaries or of uncompensated contributors.

How expansive the class of exempt organizations or activities must be to withstand constitutional assault depends upon the State’s secular aim in granting a tax exemption. If the State chose to subsidize, by means of a tax exemption, all groups that contributed to the community’s cultural, intellectual, and moral betterment, then the exemption for religious publications could be retained, provided that the exemption swept as widely as the property tax exemption we upheld in *16Walz.6 By contrast, if Texas sought to promote reflection and discussion about questions of ultimate value and the contours of a good or meaningful life, then a tax exemption would have to be available to an extended range of associations whose publications were substantially devoted to such matters; the exemption could not be reserved for publications dealing solely with religious issues, let alone restricted to publications advocating rather than criticizing religious belief or activity, without signaling an endorsement of religion that is offensive to the principles informing the Establishment Clause. See Estate of Thornton v. Caldor, Inc., 472 U. S., at 711 (O’Connor, J., concurring) (because the statute bestows an advantage on Sabbath observers “without according similar accommodation to ethical and religious beliefs and practices of other private employees,” “[t]he message conveyed is one of endorsement of a particular religious belief, to the detriment of those who do not share it”; the statute therefore “has the effect of advancing religion, and cannot withstand Establishment Clause scrutiny”); Welsh v. United States, 398 U. S., at 356-361 (Harlan, J., concurring in result) (conscientious objector status cannot be limited to those whose opposition to war has religious roots, but must extend to those whose convictions have purely moral or philosophical sources).

It is not our responsibility to specify which permissible secular objectives, if any, the State should pursue to justify a tax exemption for religious periodicals. That charge rests with the Texas Legislature. Our task, and that of the Texas courts, is rather to ensure that any scheme of exemptions *17adopted by the legislature does not have the purpose or effect of sponsoring certain religious tenets or religious belief in general. As Justice Harlan remarked: “The Court must survey meticulously the circumstances of governmental categories to eliminate, as it were, religious gerrymanders. In any particular case the critical question is whether the circumference of legislation encircles a class so broad that it can be fairly concluded that religious institutions could be thought to fall within the natural perimeter.” Walz, 397 U. S., at 696 (separate opinion). Because Texas’ sales tax exemption for periodicals promulgating the teaching of any religious sect lacks a secular objective that would justify this preference along with similar benefits for nonreligious publications or groups, and because it effectively endorses religious belief, the exemption manifestly fails this test.7

IV

A

In defense of its sales tax exemption for religious publications, Texas claims that it has a compelling interest in avoiding violations of the Free Exercise and Establishment Clauses, and that the exemption serves that end. Without such an exemption, Texas contends, its sales tax might trammel free exercise rights, as did the flat license tax this Court struck down as applied to proselytizing by Jehovah’s Witnesses in Murdock v. Pennsylvania, 319 U. S. 105 (1943). In addition, Texas argues that an exemption for religious publications neither advances nor inhibits religion, as required by the Establishment Clause, and that its elimination would entangle church and state to a greater degree than the exemption itself.

*18We reject both parts of this argument. Although Texas may widen its exemption consonant with some legitimate secular purpose, nothing in our decisions under the Free Exercise Clause prevents the State from eliminating altogether its exemption for religious publications. “It is virtually self-evident that the Free Exercise Clause does not require an exemption from a governmental program unless, at a minimum, inclusion in the program actually burdens the claimant’s freedom to exercise religious rights. ” Tony and Susan Alamo Foundation v. Secretary of Labor, 471 U. S. 290, 303 (1985) (citations omitted). In this case, the State has adduced no evidence that the payment of a sales tax by subscribers to religious periodicals or purchasers of religious books would offend their religious beliefs or inhibit religious activity. The State therefore cannot claim persuasively that its tax exemption is compelled by the Free Exercise Clause in even a single instance, let alone in every case. No concrete need to accommodate religious activity has been shown.8

*19Moreover, even if members of some religious group succeeded in demonstrating that payment of a sales tax — or, less plausibly, of a sales tax when applied to printed matter— would violate their religious tenets, it is by no means obvious that the State would be required by the Free Exercise Clause to make individualized exceptions for them. In United States v. Lee, 455 U. S. 252 (1982), we ruled unanimously that the Federal Government need not exempt an Amish employer from the payment of Social Security taxes, notwithstanding our recognition that compliance would offend his religious beliefs. We noted that “[n]ot all burdens on religion are unconstitutional,” id., at 257, and held that “[t]he state may justify a limitation on religious liberty by showing that it is essential to accomplish an overriding governmental interest.” Id., at 257-258. Although the balancing test we set forth in Lee must be performed on a case-by-case basis, a State’s interest in the uniform collection of a *20sales tax appears comparable to the Federal Government’s interest in the uniform collection of Social Security taxes, and mandatory exemptions under the Free Exercise Clause are arguably as difficult to prove. No one has suggested that members of any of the major religious denominations in the United States — the principal beneficiaries of Texas’ tax exemption — could demonstrate an infringement of their free exercise rights sufficiently serious to overcome the State’s countervailing interest in collecting its sales tax.

B

Texas’ further claim that the Establishment Clause mandates, or at least favors, its sales tax exemption for religious periodicals is equally unconvincing. Not only does the exemption seem a blatant endorsement of religion, but it appears, on its face, to produce greater state entanglement with religion than the denial of an exemption. As Justice Stevens has noted: “[There exists an] overriding interest in keeping the government — whether it be the legislature or the courts — out of the business of evaluating the relative merits of differing religious claims. The risk that governmental approval of some and disapproval of others will be perceived as favoring one religion over another is an important risk the Establishment Clause was designed to preclude.” Id., at 263, n. 2 (concurring in judgment). See Bob Jones University v. United States, 461 U. S., at 604, n. 30. The prospect of inconsistent treatment and government embroilment in controversies over religious doctrine seems especially baleful where, as in the case of Texas’ sales tax exemption, a statute requires that public officials determine whether some message or activity is consistent with “the teaching of the faith.” See, e. g., Jones v. Wolf, 443 U. S. 595 (1979); Serbian Eastern Orthodox Diocese v. Milivojevich, 426 U. S. 696 (1976); Presbyterian Church in U. S. v. Mary Elizabeth Blue Hull Memorial Presbyterian Church, 393 U. S. 440 (1969).9

*21While Texas is correct in pointing out that compliance with government regulations by religious organizations and the monitoring of their compliance by government agencies would itself enmesh the operations of church and state to some degree, we have found that such compliance would generally not impede the evangelical activities of religious groups and that the “routine and factual inquiries” commonly associated with the enforcement of tax laws “bear no resemblance to the kind of government surveillance the Court has previously held to pose an intolerable risk of government entanglement with religion.” Tony and Susan Alamo Foundation v. Secretary of Labor, 471 U. S., at 305.

On the record before us, neither the Free Exercise Clause nor the Establishment Clause prevents Texas from withdrawing its current exemption for religious publications if it chooses not to expand it to promote some legitimate secular aim.

C

Our conclusion today is admittedly in tension with some unnecessarily sweeping statements in Murdock v. Pennsylvania, 319 U. S. 105 (1943), and Follett v. McCormick, 321 U. S. 573 (1944). To the extent that language in those opinions is inconsistent with our decision here, based on the evolution in our thinking about the Religion Clauses over the last 45 years, we disavow it.

*22In Murdock, the Court ruled that a city could not impose a flat license tax payable by “all persons canvassing for or soliciting . . . orders for goods, paintings, pictures, wares, or merchandise of any kind” on Jehovah’s Witnesses who “went about from door to door . . . distributing literature and soliciting people to ‘purchase’ certain religious books and pamphlets.” 319 U. S., at 106. In Follett, the Court ruled similarly that a Jehovah’s Witness who “went from house to house distributing certain books” was exempt under the Free Exercise Clause from payment of a flat business and occupation tax on booksellers. 321 U. S., at 574. In both cases, the majority stated that the “sale” of religious pamphlets by itinerant evangelists was a form of preaching, Murdock, supra, at 109; Follett, supra, at 577, and that imposing a license or occupation tax on such a preacher was tantamount to exacting “a tax from him for the privilege of delivering a sermon.” Murdock, 319 U. S., at 112. The Court acknowledged that imposing an income or property tax on preachers would not be unconstitutional. Ibid. It emphasized, however, that a flat license or occupation tax poses a greater threat to the free exercise of religion than do those other taxes, because it is “levied and collected as a condition to the pursuit of activities whose enjoyment is guaranteed by the First Amendment” and thus “restrains in advance those constitutional liberties . . . and inevitably tends to suppress their exercise.” Id., at 114. See Follett, supra, at 575.

If one accepts the majority’s characterization of the critical issues in Murdock and Follett, those decisions are easily compatible with our holding here. In striking down application of the town ordinance to Jehovah’s Witnesses in Follett — an ordinance the Court found to be “in all material respects the same,” 321 U. S., at 574, as the one whose application it restricted in Murdock — the Court declared that only a single “narrow” question was presented: “It is whether a flat license tax as applied to one who earns his livelihood as an evangelist or preacher in his home town is constitutional.” 321 U. S., *23at 576. Regarding Follett in this light, we must agree that “we have quite a different case from that of a merchant who sells books-at a stand or on the road.” Ibid. There is no doubt that the First Amendment prevents both the States and the Federal Government from imposing a special occupation tax exclusively on those who devote their days to spreading religious messages. Moreover, it is questionable whether, consistent with the Free Exercise Clause, government may exact a facially neutral license fee designed for commercial salesmen from religious missionaries whose principal work is preaching and who only occasionally sell religious tracts for small sums, so long as “the fee is not a nominal one, imposed as a regulatory measure and calculated to defray the expense of protecting those on the streets and at home against the abuses of solicitors.” Murdock, supra, at 116. In such a case, equal treatment of commercial and religious solicitation might result in an unconstitutional imposition on religious activity warranting judicial relief, particularly where that activity is deemed central to a given faith, as the Court found this form of proselytizing to be in Murdock and Follett, and where the tax burden is far from negligible.10

*24Insofar as the Court’s holdings in Murdock and Follett are limited to these points, they are plainly consistent with our decision today.- The sales tax that Texas imposes is not an occupation tax levied on religious missionaries. Nor is it a flat tax that “restrains iii. advance,” 319 U. S., at 114, the free exercise.of religioriN' On the contrary, because the tax is equal to a small fraction of the value of each sale and payable by the buyer, it poses little danger of stamping out missionary work involving the sale of religious publications, and in view of its generality it can hardly be viewed as a covert attempt to curtail religious activity. We therefore see no inconsistency between our former decisions and our present holding.

To the extent that our opinions in Murdock and Follett might be read, however, to suggest that the States and the Federal Government may never tax the sale of religious or other publications, we reject those dicta.11 Our intervening decisions make clear that even if the denial of tax benefits “will inevitably have a substantial impact” on religious groups, the refusal to grant such benefits does not offend the Free Exercise Clause when it does not prevent those groups “from observing their religious tenets.” Bob Jones Univer*25sity v. United States, 461 U. S., at 603-604. In Murdock and Follett, the application of a flat license or occupation tax to Jehovah’s Witnesses arguably did prevent adherents of that sect from acting in accordance with some of their central religious beliefs, in the absence of any overriding government interest in denying them an exemption.12 In the much more common circumstances exemplified by this case, however, taxes or regulations would not subject religious organizations to undue burdens and the government’s interest in their uniform application is far weightier. Hence, there is no bar to Texas’ imposing a general sales tax on religious publications.

V

We conclude that Texas’ sales tax exemption for religious publications violates the First Amendment, as made applicable to the States by the Fourteenth Amendment. Accordingly, the judgment of the Texas Court of Appeals is reversed, and the case is remanded for further proceedings.

It is so ordered.

concurring in the judgment.

The Texas law at issue here discriminates on the basis of the content of publications: it provides that “[pjeriodicals . . . that consist wholly of writings promulgating the teaching of (a religious faith). . . are exempted” from the burdens of the sales tax law. Tex. Tax Code Ann. § 161.312 (1982). Thus, *26the content of a publication determines whether its publisher is exempt or nonexempt. Appellant is subject to the tax, but other publications are not because of the message they carry. This is plainly forbidden by the Press Clause of the First Amendment. Arkansas Writers’ Project, Inc. v. Ragland, 481 U. S. 221 (1987), our most recent decision to this effect, is directly applicable here, and is the proper basis for reversing the judgment below.

with whom The Chief Justice and Justice Kennedy join, dissenting.

As a judicial demolition project, today’s decision is impressive. The machinery employed by the opinions of Justice Brennan and Justice Blackmun is no more substantial than the antinomy that accommodation of religion may be required but not permitted, and the bold but unsupportable assertion (given such realities as the text of the Declaration of Independence, the national Thanksgiving Day proclaimed by every President since Lincoln, the inscriptions on our coins, the words of our Pledge of Allegiance, the invocation with *30which sessions of our Court are opened and, come to think of it, the discriminatory protection of freedom of religion in the Constitution) that government may not “convey a message of endorsement of religion.” With this frail equipment, the Court topples an exemption for religious publications of a sort that expressly appears in the laws of at least 15 of the 45 States that have sales and use taxes1 — States from Maine to Texas, from Idaho to New Jersey.2 In practice, a similar *31exemption may well exist in even more States than that, since until today our case law has suggested that it is not only permissible but perhaps required. See Follett v. McCormick, 321 U. S. 573 (1944); Murdock v. Pennsylvania, 319 U. S. 105 (1943). I expect, for example, that even in States without express exemptions many churches, and many tax assessors, have thought sales taxes inapplicable to the religious literature typically offered for sale in church foyers.

When one expands the inquiry to sales taxes on items other than publications and to other types of taxes such as property, income, amusement, and motor vehicle taxes — all of which are likewise affected by today’s holding — the Court’s accomplishment is even more impressive. At least 45 States provide exemptions for religious groups without analogous exemptions for other types of nonprofit institutions.3 For *32over half a century the federal Internal Revenue Code has allowed “ministerfs] of the gospel” (a term interpreted broadly enough to include cantors and rabbis) to exclude from gross *33income the rental value of their parsonages. 26 U. S. C. §107; see also §213(b)(ll) of the Revenue Act of 1921, ch. 136, 42 Stat. 239. In short, religious tax exemptions of the type the Court invalidates today permeate the state and federal codes, and have done so for many years.

I dissent because I find no basis in the text of the Constitution, the decisions of this Court, or the traditions of our people for disapproving this longstanding and widespread practice.

I

The opinions of Justice Brennan and Justice Black-mun proceed as though this were a matter of first impression. It is not. Nineteen years ago, in Walz v. Tax Comm’n of New York City, 397 U. S. 664 (1970), we considered and rejected an Establishment Clause challenge that was in all relevant respects identical. Since today’s opinions barely acknowledge the Court’s decision in that case (as opposed to the separate concurrences of Justices Brennan and Harlan), it requires some discussion here. Walz involved *34New York City’s grant of tax exemptions, pursuant to a state statute and a provision of the State Constitution, to “religious organizations for religious properties used solely for religious worship.” Id., at 666-667, and n. 1. In upholding the exemption, we conducted an analysis that contains the substance of the three-pronged “test” adopted the following Term in Lemon v. Kurtzman, 403 U. S. 602 (1971). First, we" concluded that “[t]he legislative purpose of the property tax exemption is neither the advancement nor the inhibition of religion.” 397 U. S., at 672. We reached that conclusion because past cases and the historical record established that property tax exemption “constitutes a reasonable and balanced attempt to guard against” the “latent dangers” of government hostility to religion. Id., at 673. We drew a distinction between an unlawful intent to favor religion and a lawful intent to “‘accommodat[e] the public service to [the people’s] spiritual needs,’” id., at 672 (quoting Zorach v. Clauson, 343 U. S. 306, 314 (1952)), and found only the latter to be involved in “sparing the exercise of religion from the burden of property taxation levied on private profit institutions,” 397 U. S., at 673.

We further concluded that the exemption did not have the primary effect of sponsoring religious activity. We noted that, although tax exemptions may have the same economic effect as state subsidies, for Establishment Clause purposes such “indirect economic benefit” is significantly different.

“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. . . . There is no genuine nexus between tax exemption and establishment of religion.” Id., at 675.

also recognized this distinction in his concurring opinion:

*35“Tax. exemptions and general subsidies, however, are qualitatively different. Though both provide economic assistance, they do so in fundamentally different ways. A subsidy involves the direct transfer of public monies to the subsidized enterprise and uses resources exacted from taxpayers as a whole. An exemption, on the other hand, involves no such transfer.” Id., at 690 (footnote omitted).

See also id., at 691 (“Tax exemptions . . . constitute mere passive state involvement with religion and not the affirmative involvement characteristic of outright governmental subsidy”).

Third, we held that the New York exemption did not produce unacceptable government entanglement with religion. In fact, quite to the contrary. Since the exemptions avoided the “tax liens, tax foreclosures, and the direct confrontations and conflicts that follow in the train of those legal processes,” id., at 674, we found that their elimination would increase government’s involvement with religious institutions, id., at 674-676. See also id., at 691 (Brennan, J., concurring) (“[I]t cannot realistically be said that termination of religious tax exemptions would quantitatively lessen the extent of state involvement with religion”).

We recognized in Walz that the exemption of religion from various taxes had existed without challenge in the law of all 50 States and the National Government before, during, and after the framing of the First Amendment’s Religion Clauses, and had achieved “undeviating acceptance” throughout the 200-year history of our Nation. “Few concepts,” we said, “are more deeply embedded in the fabric of our national life, beginning with pre-Revolutionary colonial times, than for the government to exercise at the very least this kind of benevolent neutrality toward churches and religious exercise generally so long as none was favored over others and none suffered interference.” Id., at 676-677. See also id., at 681 (Brennan, J., concurring) (noting the “the undeviating ac*36ceptance given religious tax exemptions from our earliest days as a Nation”).

It should be apparent from this discussion that Walz, which we have reaffirmed on numerous occasions in the last two decades, e. g., Corporation of Presiding Bishop of Church of Jesus Christ of Latter-day Saints v. Amos, 483 U. S. 327 (1987), is utterly dispositive of the Establishment Clause claim before us here. The Court invalidates § 151.312 of the Texas Tax Code only by distorting the holding of that case and radically altering the well-settled Establishment Clause jurisprudence which that case represents.

Justice Brennan explains away Walz by asserting that “[t]he breadth of New York’s property tax exemption was essential to our holding that it was ‘not aimed at establishing, sponsoring, or supporting religion.’” Ante, at 12 (quoting Walz, 397 U. S., at 674). This is not a plausible reading of the opinion. At the outset of its discussion concerning the permissibility of the legislative purpose, the Walz Court did discuss the fact that the New York tax exemption applied not just to religions but to certain other “nonprofit” groups, including “hospitals, libraries, playgrounds, scientific, professional, historical, and patriotic groups.” Id., at 673. The finding of valid legislative purpose was not rested upon that, however, but upon the more direct proposition that “exemption constitutes a reasonable and balanced attempt to guard against” the “latent dangers” of governmental hostility towards religion “inherent in the imposition of property taxes.” Ibid. The venerable federal legislation that the Court cited to support its holding was not legislation that exempted religion along with other things, but legislation that exempted religion alone. See, e. g., ch. 17, 6 Stat. 116 (1813) (remitting duties paid on the importation of plates for printing Bibles); ch. 91, 6 Stat. 346 (1826) (remitting duties paid on the importation of church vestments, furniture, and paintings); ch. 259, 6 Stat. 600 (1834) (remitting duties paid on the importation of church bells). Moreover, if the Court had in*37tended to rely upon a “breadth of coverage” rationale, it would have had to identify some characteristic that rationally placed religion within the same policy category as the other institutions. Justice Brennan’s concurring opinion in Walz conducted such an analysis, finding the New York exemption permissible only because religions, like the other types of nonprofit organizations exempted, “contribute to the well-being of the community in a variety of nonreligious ways,” 397 U. S., at 687, and (incomprehensibly) because they “uniquely contribute to the pluralism of American society by their religious activities,” id., at 689. (I say incomprehensibly because to favor religion for its “unique contribution” is to favor religion as religion.) Justice Harlan’s opinion conducted a similar analysis, finding that the New York statute “defined a class of nontaxable entities whose common denominator is their nonprofit pursuit of activities devoted to cultural and moral improvement and the doing of ‘good works’ by performing certain social services in the community that might otherwise have to be assumed by government.” Id., at 696. The Court’s opinion in Walz, however, not only failed to conduct such an analysis, but — seemingly in reply to the concurrences — explicitly and categorically disavowed reliance upon it, concluding its discussion of legislative purpose with a paragraph that begins as follows: “We find it unnecessary to justify the tax exemption on the social welfare services or ‘good works’ that some churches perform for parishioners and others,” id., at 674. This should be compared with today’s rewriting of Walz: “[W]e concluded that the State might reasonably have determined that religious groups generally contribute to the cultural and moral improvement of the community, perform useful social services, and enhance a desirable pluralism of viewpoint and enterprise, just as do the host of other nonprofit organizations that qualified for the exemption.” Ante, at 12, n. 2. This is a marvellously accurate description of what Justices Brennan and Harlan believed, and what the Court specifically re*38jected. The Court did not approve an exemption for charities that happened to benefit religion; it approved an exemption for religion as an exemption for religion.

Today’s opinions go beyond misdescribing Walz, however. In repudiating what Walz in fact approved, they achieve a revolution in bur Establishment Clause jurisprudence, effectively overruling other cases that were based, as Walz was, on the “accommodation of religion” rationale. According to Justice Brennan’s opinion, no law is constitutional whose “benefits [are] confined to religious organizations,” ante, at 11 — except, of course, those laws that are unconstitutional unless they contain benefits confined to religious organizations, see ante, at 17-18. See also Justice Blackmun’s opinion, ante, at 28. Our jurisprudence affords no support for this unlikely proposition. Walz is just one of a long line of cases in which we have recognized that “the government may (and sometimes must) accommodate religious practices and that it may do so without violating the - Establishment Clause.” Hobbie v. Unemployment Appeals Comm’n of Fla., 480 U. S. 136, 144-145 (1987); see McConnell, Accommodation of Religion, 1985 S. Ct. Rev. 1, 3. In such cases as Sherbert v. Verner, 374 U. S. 398 (1963), Wisconsin v. Yoder, 406 U. S. 205 (1972), Thomas v. Review Bd. of Ind. Employment Security Div., 450 U. S. 707 (1981), and Hobbie v. Unemployment Appeals Comm’n of Fla., supra, we held that the Free Exercise Clause of the First Amendment required religious beliefs to be accommodated by granting religion-specific exemptions from otherwise applicable laws. We have often made clear, however, that “[t]he limits of permissible state accommodation to religion are by no means coextensive with the noninterference mandated by the Free Exercise Clause.” Walz, 397 U. S., at 673. See also Hobbie, supra, at 144-145, and n. 10; Gillette v. United States, 401 U. S. 437, 453 (1971); Braunfeld v. Brown, 366 U. S. 599, 605-608 (1961) (plurality opinion); Wallace v. Jaffree, 472 U. S. 38, 82 (1985) (O’Connor, J., concurring).

*39We applied the accommodation principle, to permit special treatment of religion that was not required by the Free Exercise Clause, in Zorach v. Clauson, 343 U. S. 306 (1952), where we found no constitutional objection to a New York City program permitting public school children to absent themselves one hour a week for “religious observance and education outside the school grounds,” id., at 308, n. 1. We applied the same principle only two Terms ago in Corporation of Presiding Bishop, where, citing Zorach and Walz, we upheld a section of the Civil Rights Act of 1964 exempting religious groups (and only religious groups) from Title VII’s antidiscrimination provisions. We found that “it is a permissible legislative purpose to alleviate significant governmental interference with the ability of religious organizations to define and carry out their religious missions.” 483 U. S., at 335. We specifically rejected the District Court’s conclusion identical to that which a majority of the Court endorses today: that invalidity followed from the fact that the exemption “singles out religious entities for a benefit, rather than benefiting a broad grouping of which religious organizations are only a part.” Id., at 333. We stated that the Court “has never indicated that statutes that give special consideration to religious groups are per se invalid.” Id., at 338. As discussed earlier, it was this same principle of permissible accommodation that we applied in Walz.

The novelty of today’s holding is obscured by Justice Brennan’s citation and description of many eases in which “breadth of coverage” ivas relevant to the First Amendment determination. See ante, at 10-11. Breadth of coverage is essential to constitutionality whenever a law’s benefiting of religious activity is sought to be defended not specifically (or not exclusively) as an intentional and reasonable accommodation of religion, but as merely the incidental consequence of seeking to benefit all activity that achieves a particular secular goal. But that is a different rationale — more commonly invoked than accommodation of religion but, as our cases *40show, not preclusive of it. Where accommodation of religion is the justification, by definition religion is being singled out. The same confusion of rationales explains the facility with which Justice Brennan’s opinion can portray the present statute as violating the first prong of the Lemon test, which is usually described as requiring a “secular legislative purpose.” Lemon, 403 U. S., at 612. That is an entirely accurate description of the governing rule when, as in Lemon and most other cases, government aid to religious institutions is sought to be justified on the ground that it is not religion per se that is the object of assistance, but rather the secular functions that the religious institutions, along with other institutions, provide. But as I noted earlier, the substance of the Lemon test (purpose, effect, entanglement) was first roughly set forth in Walz — and in that context, the “accommodation of religion” context, the purpose was said to be valid so long as it was “neither the advancement nor the inhibition of religion; . . . neither sponsorship nor hostility.” 397 U. S., at 672. Of course rather than reformulating the Lemon test in “accommodation” cases (the text of Lemon is not, after all, a statutory enactment), one might instead simply describe the protection of free exercise concerns, and the maintenance of the necessary neutrality, as “secular purpose and effect,” since they are a purpose and effect approved, and indeed to some degree mandated, by the Constitution. However the reconciliation with the Lemon terminology is achieved, our cases make plain that it is permissible for a State to act with the purpose and effect of “limiting governmental interference with the exercise of religion.” Corporation of Presiding Bishop, 483 U. S., at 339.

It is not always easy to determine when accommodation slides over into promotion, and neutrality into favoritism, but the withholding of a tax upon the dissemination of religious materials is not even a close case. The subjects of the exemption before us consist exclusively of “writings promulgating the teaching of the faith” and “writings sacred to a reli*41gious faith.” If there is any close question, it is not whether the exemption is permitted, but whether it is constitutionally compelled in order to avoid “interference with the dissemination of religious ideas.” Gillette, 401 U. S., at 462. In Murdock v. Pennsylvania, 319 U. S. 105 (1943), we held that it was unconstitutional to apply a municipal license tax on door-to-door solicitation to sellers of religious books and pamphlets. One Term later, in Follett v. McCormick, 321 U. S. 573 (1944), we held that it was unconstitutional to apply to such persons a municipal license tax on “[ajgents selling books.” Those cases are not as readily distinguishable as Justice Brennan suggests. I doubt whether it would have made any difference (as he contends) if the municipalities had attempted to achieve the same result of burdening the religious activity through a sales tax rather than a license tax; surely such a distinction trivializes the holdings. And the other basis of distinction he proposes —that the persons taxed in those cases were “religious missionaries whose principal work is preaching” — is simply not available with respect to the first part of the statute at issue here (which happens to be the portion upon which petitioner placed its exclusive reliance). Unlike the Texas exemption for sacred books, which, on its face at least, applies to all sales, the exemption for periodicals applies to material that not only “consist[s] wholly of writings promulgating the teaching of [a religious] faith,” but also is “published or distributed by [that] faith.” Surely this is material distributed by missionaries. Unless, again, one wishes to trivialize the earlier cases, whether they are full-time or part-time missionaries can hardly make a difference, nor can the fact that they conduct their proselytizing through the mail or from a church or store instead of door-to-door.

I am willing to acknowledge, however, that Murdock and Follett are narrowly distinguishable. But what follows from that is not the facile conclusion that therefore the State has no “compelling interest in avoiding violations of the Free Ex*42ercise and Establishment Clauses,” ante, at 17, and thus the exemption is invalid. This analysis is yet another expression of Justice Brennan’s repudiation of the accommodation principle — which, as described earlier, consists of recognition that “[t]he limits of permissible state accommodation to religion are by no means co-extensive with the noninterference mandated by the Free Exercise Clause.” Walz, 397 U. S., at 673. By saying that what is not required cannot be allowed, Justice Brennan would completely block off the already narrow “channel between the Scylla [of what the Free Exercise Clause demands] and the Charybdis [of what the Establishment Clause forbids] through which any state or federal action must pass in order to survive constitutional scrutiny.” Thomas, 450 U. S., at 721 (Rehnquist, J., dissenting). The proper lesson to be drawn from the narrow distinguishing of Murdock and Follett is quite different: If the exemption comes so close to being a constitutionally required accommodation, there is no doubt that it is at least a permissible one.

Although Justice Brennan’s opinion places almost its entire reliance upon the “purpose” prong of Lemon, it alludes briefly to the second prong as well, finding that § 151.312 has the impermissible “effect of sponsoring certain religious tenets or religious belief in general,” ante, at 17. Once again, Walz stands in stark opposition to this assertion, but it may be useful to explain why. Quite obviously, a sales tax exemption aids religion, since it makes it less costly for religions to disseminate their beliefs. Cf. Murdock, supra, at 112-113. But that has never been enough to strike down an enactment under the Establishment Clause. “A law is not unconstitutional simply because it allows churches to advance religion, which is their very purpose.” Corporation of Presiding Bishop, supra, at 337 (emphasis in original). The Court has consistently rejected “the argument that any program which in some manner aids an institution with a religious affiliation” violates the Establishment Clause. Muel*43ler v. Allen, 463 U. S. 388, 393 (1983) (quoting Hunt v. McNair, 413 U. S. 734, 742 (1973)). To be sure, we have set our face against the subsidizing of religion — and in other contexts we have suggested that tax exemptions and subsidies are equivalent. E. g., Bob Jones University v. United States, 461 U. S. 574, 591 (1983); Regan v. Taxation With Representation of Wash., 461 U. S. 540, 544 (1983). We have not treated them as equivalent, however, in the Establishment Clause context, and with good reason. “In the case of direct subsidy, the state forcibly diverts the income of both believers and nonbelievers to churches. In the case of an exemption, the state merely refrains from diverting to its own uses income independently generated by the churches through voluntary contributions.” Giannella, Religious Liberty, Nonestablishment, and Doctrinal Development, 81 Harv. L. Rev. 513, 553 (1968). In Walz we pointed out that the primary effect of a tax exemption was not to sponsor religious activity but to “restric[t] the fiscal relationship between church and state” and to “complement and reinforce the desired separation insulating each from the other.” 397 U. S., at 676; see also id., at 690-691 (Brennan, J., concurring).

Finally, and least persuasively of all, Justice Brennan suggests that § 151.312 violates the “excessive government entanglement” aspect of Lemon, 403 U. S., at 613. Ante, at 20-21. It is plain that the exemption does not foster the sort of “comprehensive, discriminating, and continuing state surveillance” necessary to run afoul of that test. 403 U. S., at 619. A State does not excessively involve itself in religious affairs merely by examining material to determine whether it is religious or secular in nature. Mueller v. Allen, supra, at 403; Meek v. Pittenger, 421 U. S. 349, 359-362 (1975) (upholding loans of nonreligious textbooks to religious schools); Board of Education of Central School Dist. No. 1 v. Allen, 392 U. S. 236 (1968) (same). In Mueller, for instance, we held that state officials’ examination of textbooks to determine whether they were “books and materials used in the *44teaching of religious tenets, doctrines or worship” did not constitute excessive entanglement. 463 U. S., at 403. I see no material distinction between that inquiry and the one Texas officials must make in this case. Moreover, here as in Walz, see 397 U. S., at 674, it is all but certain that elimination of the exemption will have the effect of increasing government’s involvement with religion. The Court’s invalidation of § 151.312 ensures that Texas churches selling publications that promulgate their religion will now be subject to numerous statutory and regulatory impositions, including audits, Tex. Tax Code Ann. § 151.023 (1982 and Supp. 1988-1989), requirements for the filing of security, § 151.251 et seq., reporting requirements, §151.401 et seq., writs of attachment without bond, §151.605, tax liens, §151.608, and the seizure and sale of property to satisfy tax delinquencies, §151.610.

II

Having found that this statute does not violate the Establishment Clause of the First Amendment, I must consider whether it violates the Press Clause, pursuant to our decision two Terms ago in Arkansas Writers’ Project, Inc. v. Ragland, 481 U. S. 221 (1987). Although I dissented in Ragland, even accepting it to be correct I cannot conclude as readily as does Justice White, ante, at 26, that it applies here.

The tax exemption at issue in Ragland, which we held to be unconstitutional because rontent based, applied to trade publications and sports magazines along with religious periodicals and sacred writings, and hence could not be justified as an accommodation of religion. If the purpose of accommodating religion can support action that might otherwise violate the Establishment Clause, I see no reason why it does not also support action that might otherwise violate the Press Clause or the Speech Clause. To hold otherwise would be to narrow the accommodation principle enormously, leaving it applicable to only nonexpressive religious worship. I do not *45think that is the law. Just as the Constitution sometimes requires accommodation of religious expression despite not only the Establishment Clause but also the Speech and Press Clauses, so also it sometimes ■permits accommodation despite all those Clauses. Such accommodation is unavoidably content based — because the Freedom of Religion Clause is content based.

It is absurd to think that a State which chooses to prohibit booksellers from making stories about seduction available to children of tender years cannot make an exception for stories contained in sacred writings (e. g., the story of Susanna and the Two Elders, Daniel 13:1-65). And it is beyond imagination that the sort of tax exemption permitted (indeed, required) by Murdock and Follett would have to be withdrawn if door-to-door salesmen of commercial magazines demanded equal treatment with Seventh-day Adventists on Press Clause grounds. And it is impossible to believe that the State is constitutionally prohibited from taxing Texas Monthly magazine more heavily than the Holy Bible.

* * *

Today’s decision introduces a new strain of irrationality in our Religion Clause jurisprudence. I have no idea how to reconcile it with Zorach (which seems a much harder case of accommodation), with Walz (which seems precisely in point), and with Corporation of Presiding Bishop (on which the ink is hardly dry). It is not right — it is not constitutionally healthy — that this Court should feel authorized to refashion anew our civil society’s relationship with religion, adopting a theory of church and state that is contradicted by current practice, tradition, and even our own case law. I dissent.