Opinion

JUSTICE O'CONNOR, concurring in part and concurring in the judgment.

I

The question at the heart of these cases is: What may the government do, consistently with the Establishment Clause, to accommodate people's religious beliefs? The history of the Satmars in Orange County is especially instructive on this, because they have been involved in at least three accommodation problems, of which these cases are only the most recent.

The first problem related to zoning law, and arose shortly after the Satmars moved to the town of Monroe in the early 1970's. Though the area in which they lived was zoned for single-family homes, the Satmars subdivided their houses into several apartments, apparently in part because of their traditionally close-knit extended family groups. The Satmars also used basements of some of their buildings as schools and synagogues, which according to the town was also a zoning violation. See N. Y. Times, Oct. 17, 1976, section 1, p. 53, col. 1; App. 10-14.

Fortunately for the Satmars, New York state law had a way of accommodating their concerns. New York allows virtually any group of residents to incorporate their own village, with broad powers of self-government. The Satmars followed this course, incorporating their community as the village of Kiryas Joel, and their zoning problems, at least, were solved. Ante, at 691.

The Satmars' next need for accommodation arose in the mid-1980's. Satmar education is pervasively religious, and is provided through entirely private schooling. But though the Satmars could afford to educate most of their children, educating the handicapped is a difficult and expensive business. Moreover, it is a business that the government generally funds, with tax moneys that come from the Satmars as well as from everyone else. In 1984, therefore, the MonroeWoodbury Central School District began providing handicapped education services to the Satmar children at an annex to the Satmar religious school. The curriculum and the environment of the services were entirely secular. They were the same sort of services available to handicapped students at secular public and private schools throughout the country.

In 1985, however, we held that publicly funded classes on religious school premises violate the Establishment Clause. School Dist. of Grand Rapids v. Ball, 473 U. S. 373 ; Aguilar v. Felton, 473 U. S. 402. Based on these decisions, the Monroe-Woodbury Central School District stopped providing services at the Kiryas Joel site, and required the Satmar children to attend public schools outside the village. This, however, was not a satisfactory arrangement for the Satmars, in part because the Satmar children had a hard time dealing with immersion in the non-Satmar world. By 1989, only one handicapped Kiryas Joel child was going to the public school-the others were getting either privately funded services or no special education at all. Though the Satmars tried to reach some other arrangement with the MonroeWoodbury Central School District, the problem was not resolved.

In response to these difficulties came the third accommodation. In 1989, the New York Legislature passed a statute to create a special school district covering only the village of Kiryas Joel. This school district could, of course, only operate secular schools, and the Satmars therefore wanted to use it only to provide education for the handicapped. But because the district provides this education in the village, Satmar children could take advantage of the district's serv ices without encountering the problems they faced when they were sent out to Monroe-Woodbury schools. It is the constitutionality of the law creating this district that we are now called on to decide.

II

The three situations outlined above shed light on an important aspect of accommodation under the First Amendment:

Religious needs can be accommodated through laws that are neutral with regard to religion. The Satmars' living arrangements were accommodated by their right-a right shared with all other communities, religious or not, throughout New York-to incorporate themselves as a village. From 1984 to 1985, the Satmar handicapped children's educational needs were accommodated by special education programs like those available to all handicapped children, religious or not. Other examples of such accommodations abound: The Constitution itself, for instance, accommodates the religious desires of those who were opposed to oaths by allowing any officeholder-of any religion, or none-to take either an oath of office or an affirmation. Art. II, § 1, cl. 8; Art. VI, cl. 3; see also Arndt. 4. Likewise, the selective service laws provide exemptions for conscientious objectors whether or not the objection is based on religious beliefs. Welsh v. United States, 398 U. S. 333, 356 (1970) (Harlan, J., concurring in result).

We have time and again held that the government generally may not treat people differently based on the God or gods they worship, or do not worship. "The clearest command of the Establishment Clause is that one religious denomination cannot be officially preferred over another." Larson v. Valente, 456 U. S. 228, 244 (1982). "Just as we subject to the most exacting scrutiny laws that make classifications based on race... so too we strictly scrutinize governmental classifications based on religion." Employment Div., Dept. of Human Resources of Ore. v. Smith, 494 U. S. 872, 886, n. 3 (1990). "[T]he Establishment Clause prohibits government from abandoning secular purposes... to favor the adherents of any sect or religious organization." Gil lette v. United States, 401 U. S. 437, 450 (1971). "Neither [the State nor the Federal Governments] can constitutionally pass laws or impose requirements which aid all religions as against non-believers, and neither can aid those religions based on a belief in the existence of God as against those religions founded on different beliefs." Torcaso v. Watkins, 367 U. S. 488, 495 (1961) (footnote omitted). See also Texas Monthly, Inc. v. Bullock, 489 U. S. 1, 8-9 (1989) (plurality opinion); id., at 26, 28-29 (BLACKMUN, J., concurring in judgment); Welsh, supra, at 356 (Harlan, J., concurring); Walz v. Tax Comm'n of City of New York, 397 U. S. 664, 696-697 (1970) (opinion of Harlan, J.).

This emphasis on equal treatment is, I think, an eminently sound approach. In my view, the Religion Clauses-the Free Exercise Clause, the Establishment Clause, the Religious Test Clause, Art. VI, cl. 3, and the Equal Protection Clause as applied to religion-all speak with one voice on this point: Absent the most unusual circumstances, one's religion ought not affect one's legal rights or duties or benefits. As I have previously noted, "the Establishment Clause is infringed when the government makes adherence to religion relevant to a person's standing in the political community." Wallace v. Jaffree, 472 U. S. 38, 69 (1985) (opinion concurring in judgment).

That the government is acting to accommodate religion should generally not change this analysis. What makes accommodation permissible, even praiseworthy, is not that the government is making life easier for some particular religious group as such. Rather, it is that the government is accommodating a deeply held belief. Accommodations may thus justify treating those who share this belief differently from those who do not; but they do not justify discriminations based on sect. A state law prohibiting the consumption of alcohol may exempt sacramental wines, but it may not exempt sacramental wine use by Catholics but not by Jews. A draft law may exempt conscientious objectors, but it may not exempt conscientious objectors whose objections are based on theistic belief (such as Quakers) as opposed to nontheistic belief (such as Buddhists) or atheistic belief. See Welsh, 398 U. S., at 356 (Harlan, J., concurring in result); see also id., at 335-344 (reaching this result on statutory interpretation grounds); United States v. Seeger, 380 U. S. 163 (1965) (same). The Constitution permits "nondiscriminatory religious-practice exemption[s]," Smith, supra, at 890 (emphasis added), not sectarian ones.

III

I join Parts I, II-B, II-C, and III of the Court's opinion because I think this law, rather than being a general accommodation, singles out a particular religious group for favorable treatment. The Court's analysis of the history of this law and of the surrounding statutory scheme, ante, at 699701, persuades me of this.

On its face, this statute benefits one group-the residents of Kiryas Joel. Because this benefit was given to this group based on its religion, it seems proper to treat it as a legislatively drawn religious classification. I realize this is a close question, because the Satmars may be the only group who currently need this particular accommodation. The legislature may well be acting without any favoritism, so that if another group came to ask for a similar district, the group might get it on the same terms as the Satmars. But the nature of the legislative process makes it impossible to be sure of this. A legislature, unlike the judiciary or many administrative decisionmakers, has no obligation to respond to any group's requests. A group petitioning for a law may never get a definite response, or may get a "no" based not on the merits but on the press of other business or the lack of an influential sponsor. Such a legislative refusal to act would not normally be reviewable by a court. Under these circumstances, it seems dangerous to validate what appears to me a clear religious preference.

Our invalidation of this statute in no way means that the Satmars' needs cannot be accommodated. There is nothing improper about a legislative intention to accommodate a religious group, so long as it is implemented through generally applicable legislation. New York may, for instance, allow all villages to operate their own school districts. If it does not want to act so broadly, it may set forth neutral criteria that a village must meet to have a school district of its own; these criteria can then be applied by a state agency, and the decision would then be reviewable by the judiciary. A district created under a generally applicable scheme would be acceptable even though it coincides with a village that was consciously created by its voters as an enclave for their religious group. I do not think the Court's opinion holds the contrary.

I also think there is one other accommodation that would be entirely permissible: the 1984 scheme, which was discontinued because of our decision in Aguilar. The Religion Clauses prohibit the government from favoring religion, but they provide no warrant for discriminating against religion. All handicapped children are entitled by law to governmentfunded special education. See, e. g., Individuals with Disabilities Education Act, 20 U. S. C. § 1400 et seq. If the government provides this education on-site at public schools and at nonsectarian private schools, it is only fair that it provide it on-site at sectarian schools as well.

I thought this to be true in Aguilar, see 473 U. S., at 421431 (dissenting opinion), and I still believe it today. The Establishment Clause does not demand hostility to religion, religious ideas, religious people, or religious schools. Cf. Lamb's Chapel v. Center Moriches Union Free School Dist., 508 U. S. 384 (1993). It is the Court's insistence on disfavoring religion in Aguilar that led New York to favor it here. The Court should, in a proper case, be prepared to reconsider Aguilar, in order to bring our Establishment Clause juris prudence back to what I think is the proper track-government impartiality, not animosity, toward religion.

IV

One aspect of the Court's opinion in these cases is worth noting: Like the opinions in two recent cases, Lee v. Weisman, 505 U. S. 577 (1992); Zobrest v. Catalina Foothills School Dist., 509 U. S. 1 (1993), and the case I think is most relevant to these, Larson v. Valente, 456 U. S. 228 (1982), the Court's opinion does not focus on the Establishment Clause test we set forth in Lemon v. Kurtzman, 403 U. S. 602 (1971).

It is always appealing to look for a single test, a Grand Unified Theory that would resolve all the cases that may arise under a particular Clause. There is, after all, only one Establishment Clause, one Free Speech Clause, one Fourth Amendment, one Equal Protection Clause. See Craig v. Boren, 429 U. S. 190, 211 (1976) (STEVENS, J., concurring).

But the same constitutional principle may operate very differently in different contexts. We have, for instance, no one Free Speech Clause test. We have different tests for content-based speech restrictions, for content-neutral speech restrictions, for restrictions imposed by the government acting as employer, for restrictions in nonpublic fora, and so on. This simply reflects the necessary recognition that the interests relevant to the Free Speech Clause inquiry-personal liberty, an informed citizenry, government efficiency, public order, and so on-are present in different degrees in each context.

And setting forth a unitary test for a broad set of cases may sometimes do more harm than good. Any test that must deal with widely disparate situations risks being so vague as to be useless. I suppose one can say that the general test for all free speech cases is "a regulation is valid if the interests asserted by the government are stronger than the interests of the speaker and the listeners," but this would hardly be a serviceable formulation. Similarly, Lemon has, with some justification, been criticized on this score.

Moreover, shoehorning new problems into a test that does not reflect the special concerns raised by those problems tends to deform the language of the test. Relatively simple phrases like "primary effect... that neither advances nor inhibits religion" and" 'entanglement,'" Lemon, supra, at 612-613, acquire more and more complicated definitions which stray ever further from their literal meaning. Distinctions are drawn between statutes whose effect is to advance religion and statutes whose effect is to allow religious organizations to advance religion. See, e. g., Corporation of Presiding Bishop of Church of Jesus Christ of Latter-day Saints v. Amos, 483 U. S. 327, 336-337 (1987); id., at 347 (O'CONNOR, J., concurring in judgment) (discussing this point). Assertions are made that authorizing churches to veto liquor sales in surrounding areas "can be seen as having a 'primary' and 'principal' effect of advancing religion." Larkin v. Grendel's Den, Inc., 459 U. S. 116, 125-126 (1982). "[E]ntanglement" is discovered in public employers monitoring the performance of public employees-surely a proper enough function-on parochial school premises, and in the public employees cooperating with the school on class scheduling and other administrative details. Aguilar v. Felton, 473 U. S., at 413. Alternatives to Lemon suffer from a similar failing when they lead us to find "coercive pressure" to pray when a school asks listeners-with no threat of legal sanctions-to stand or remain silent during a graduation prayer. Lee v. Weisman, supra, at 592. Some of the results and perhaps even some of the reasoning in these cases may have been right. I joined two of the cases cited above, Larkin and Lee, and continue to believe they were correctly decided. But I think it is more useful to recognize the relevant concerns in each case on their own terms, rather than trying to squeeze them into language that does not really apply to them. Finally, another danger to keep in mind is that the bad test may drive out the good. Rather than taking the opportunity to derive narrower, more precise tests from the case law, courts tend to continually try to patch up the broad test, making it more and more amorphous and distorted. This, I am afraid, has happened with Lemon.

Experience proves that the Establishment Clause, like the Free Speech Clause, cannot easily be reduced to a single test. There are different categories of Establishment Clause cases, which may call for different approaches. Some cases, like these, involve government actions targeted at particular individuals or groups, imposing special duties or giving special benefits. Cases involving government speech on religious topics, see, e. g., Lee v. Weisman, supra; Allegheny County v. American Civil Liberties Union, Greater Pittsburgh Chapter, 492 U. S. 573 (1989); Lynch v. Donnelly, 465 U. S. 668 (1984); Stone v. Graham, 449 U. S. 39 (1980), seem to me to fall into a different category and to require an analysis focusing on whether the speech endorses or disapproves of religion, rather than on whether the government action is neutral with regard to religion. See Allegheny County, supra, at 623-637 (O'CONNOR, J., concurring in part and concurring in judgment).

Another category encompasses cases in which the government must make decisions about matters of religious doctrine and religious law. See Serbian Eastern Orthodox Diocese for United States and Canada v. Milivojevich, 426 U. S. 696 (1976) (which also did not apply Lemon). These cases, which often arise in the application of otherwise neutral property or contract principles to religious institutions, involve complicated questions not present in other situations. See, e. g., 426 U. S., at 721 (looking at some aspects of religious law to determine the structure of the church, but refusing to look further into religious law to resolve the ultimate dispute). Government delegations of power to religious bodies may make up yet another category. As Larkin itself suggested, government impartiality towards religion may not be enough in such situations: A law that bars all alcohol sales within some distance of a church, school, or hospital may be valid, but an equally evenhanded law that gives each institution discretionary power over the sales may not be. Larkin, supra, at 123-124. Of course, there may well be additional categories, or more opportune places to draw the lines between the categories.

As the Court's opinion today shows, the slide away from Lemon's unitary approach is well under way. A return to Lemon, even if possible, would likely be futile, regardless of where one stands on the substantive Establishment Clause questions. I think a less unitary approach provides a better structure for analysis. If each test covers a narrower and more homogeneous area, the tests may be more precise and therefore easier to apply. There may be more opportunity to pay attention to the specific nuances of each area. There might also be, I hope, more consensus on each of the narrow tests than there has been on a broad test. And abandoning the Lemon framework need not mean abandoning some of the insights that the test reflected, nor the insights of the cases that applied it.

Perhaps eventually under this structure we may indeed distill a unified, or at least a more unified, Establishment Clause test from the cases. Cf. Clark v. Community for Creative Non-Violence, 468 U. S. 288, 298-299 (1984) (uniting two strands of Free Speech Clause doctrine). But it seems to me that the case law will better be able to evolve towards this if it is freed from the Lemon test's rigid influence. The hard questions would, of course, still have to be asked; but they will be asked within a more carefully tailored and less distorted framework.

***

For the reasons stated, I would affirm the judgment of the Court of Appeals of the State of New York.