delivered the opinion of the Court.

The City of New York uses federal funds to pay the salaries of public employees who teach in parochial schools. In this companion case to School District of Grand Rapids v. Ball, ante, p. 373, we determine whether this practice violates the Establishment Clause of the First Amendment.

hH

<C

The program at issue in this case, originally enacted as Title I of the Elementary and Secondary Education Act of 19651 authorizes the Secretary of Education to distribute financial assistance to local educational institutions to meet the needs of educationally deprived children from low-income families. The funds are to be appropriated in accordance with programs proposed by local educational agencies and approved by state educational agencies. 20 U. S. C. *405§ 3805(a).2 “To the extent consistent with the number of educationally deprived children in the school district of the local educational agency who are enrolled in private elementary and secondary schools, such agency shall make provisions for including special educational services and arrangements ... in which such children can participate.” § 3806(a).3 The proposed programs must also meet the following statutory requirements: the children involved in the program must be educationally deprived, § 3804(a),4 the children must reside in areas comprising a high concentration of low-income families, § 3805(b),5 and the programs must sup*406plement, not supplant, programs that would exist absent funding under Title I. § 3807(b).6

Since 1966, the City of New York has provided instructional services funded by Title I to parochial school students on the premises of parochial schools. Of those students eligible to receive funds in 1981-1982, 13.2% were enrolled in private schools. Of that group, 84% were enrolled in schools affiliated with the Roman Catholic Archdiocese of New York and the Diocese of Brooklyn and 8% were enrolled in Hebrew day schools. With respect to the religious atmosphere of these schools, the Court of Appeals concluded that “the picture that emerges is of a system in which religious considerations play a key role in the selection of students and teachers, and which has as its substantial purpose the inculcation of religious values.” 739 F. 2d 48, 68 (CA2 1984).

The programs conducted at these schools include remedial reading, reading skills, remedial mathematics, English as a second language, and guidance services. These programs are carried out by regular employees of the public schools (teachers, guidance counselors, psychologists, psychiatrists, and social workers) who have volunteered to teach in the parochial schools. The amount of time that each professional spends in the parochial school is determined by the number of students in the particular program and the needs of these students.

The City’s Bureau of Nonpublic School Reimbursement makes teacher assignments, and the instructors are super*407vised by field personnel, who attempt to pay at least one unannounced visit per month. The field supervisors, in turn, report to program coordinators, who also pay occasional unannounced supervisory visits to monitor Title I classes in the parochial schools. The professionals involved in the program are directed to avoid involvement with religious activities that are conducted within the private schools and to bar religious materials in their classrooms. All material and equipment used in the programs funded under Title I are supplied by the Government and are used only in those programs. The professional personnel are solely responsible for the selection of the students. Additionally, the professionals are informed that contact with private school personnel should be kept to a minimum. Finally, the administrators of the parochial schools are required to clear the classrooms used by the public school personnel of all religious symbols.

B

In 1978, six taxpayers commenced this action in the District Court for the Eastern District of New York, alleging that the Title I program administered by the City of New York violates the Establishment Clause. These taxpayers, appellees in today’s case, sought to enjoin the further distribution of funds to programs involving instruction on the premises of parochial schools. Initially the case was held for the outcome of National Coalition for Public Education and Religious Liberty v. Harris, 489 F. Supp. 1248 (SDNY 1980) (PEARL), which involved an identical challenge to the Title I program. When the District Court in PEARL affirmed the constitutionality of the Title I program, ibid., and this Court dismissed the appeal for want of jurisdiction, 449 U. S. 808 (1980), the challenge of the present appellees was renewed. The District Court granted appellants’ motion for summary judgment based upon the evidentiary record developed in PEARL.

*408A unanimous panel of the Court of Appeals for the Second Circuit reversed, holding that

“[t]he Establishment Clause, as it has been interpreted by the Supreme Court in Public Funds for Public Schools v. Marburger, 358 F. Supp. 29 (D. N. J. 1973), aff’d mem., 417 U. S. 961 . . . (1974); Meek v. Pittenger, 421 U. S. 349 . . . (1975) (particularly Part V, pp. 367-72); and Wolman v. Walter, 433 U. S. 229 . . . (1977), constitutes an insurmountable barrier to the use of federal funds to send public school teachers and other professionals into religious schools to carry on instruction, remedial or otherwise, or to provide clinical and guidance services of the sort at issue here.” 739 F. 2d, at 49-50.

We postponed probable jurisdiction. 469 U. S. 878 (1984). We conclude that jurisdiction by appeal does not properly lie.7 Treating the papers as a petition for a writ of certiorari, see 28 U. S. C. § 2103, we grant the petition and now affirm the judgment below.

II

In School District of Grand Rapids v. Ball, ante, p. 373, the Court has today held unconstitutional under the Establishment Clause two remedial and enhancement programs operated by the Grand Rapids Public School District, in which *409classes were provided to private school children at public expense in classrooms located in and leased from the local private schools. The New York City programs challenged in this case are very similar to the programs we examined in Ball. In both cases, publicly funded instructors teach classes composed exclusively of private school students in private school buildings. In both cases, an overwhelming number of the participating private schools are religiously affiliated. In both cases, the publicly funded programs provide not only professional personnel, but also all materials and supplies necessary for the operation of the programs. Finally, the instructors in both cases are told that they are public school employees under the sole control of the public school system.

Appellants attempt to distinguish this case on the ground that the City of New York, unlike the Grand Rapids Public School District, has adopted a system for monitoring the religious content of publicly funded Title I classes in the religious schools. At best, the supervision in this case would assist in preventing the Title I program from being used, intentionally or unwittingly, to inculcate the religious beliefs of the surrounding parochial school. But appellants’ argument fails in any event, because the supervisory system established by the City of New York inevitably results in the excessive entanglement of church and state, an Establishment Clause concern distinct from that addressed by the effects doctrine. Even where state aid to parochial institutions does not have the primary effect of advancing religion, the provision of such aid may nonetheless violate the Establishment Clause owing to the nature of the interaction of church and state in the administration of that aid.

The principle that the state should not become too closely entangled with the church in the administration of assistance is rooted in two concerns. When the state becomes enmeshed with a given denomination in matters of religious significance, the freedom of religious belief of those who are not adherents of that denomination suffers, even when the *410governmental purpose underlying the involvement is largely secular. In addition, the freedom of even the adherents of the denomination is limited by the governmental intrusion into sacred matters. “[T]he First Amendment rests upon the premise that both religion and government can best work to achieve their lofty aims if each is left free from the other within its respective sphere.” McCollum v. Board of Education, 333 U. S. 203, 212 (1948).

In Lemon v. Kurtzman, 403 U. S. 602 (1971), the Court held that the supervision necessary to ensure that teachers in parochial schools were not conveying religious messages to their students would constitute the excessive entanglement of church and state:

“A comprehensive, discriminating, and continuing state surveillance will inevitably be required to ensure that these restrictions are obeyed and the First Amendment otherwise respected. Unlike a book, a teacher cannot be inspected once so as to determine the extent and intent of his or her personal beliefs and subjective acceptance of the limitations imposed by the First Amendment. These prophylactic contacts will involve excessive and enduring entanglement between state and church. ” Id., at 619.

Similarly, in Meek v. Pittenger, 421 U. S. 349 (1975), we invalidated a state program that offered, inter alia, guidance, testing, and remedial and therapeutic services performed by public employees on the premises of the parochial schools. Id., at 352-353. As in Lemon, we observed that though a comprehensive system of supervision might conceivably prevent teachers from having the primary effect of advancing religion, such a system would inevitably lead to an unconstitutional administrative entanglement between church and state.

“The prophylactic contacts required to ensure that teachers play a strictly nonideological role, the Court held [in Lemon], necessarily give rise to a constitution*411ally intolerable degree of entanglement between church and state. Id., at 619. The same excessive entanglement would be required for Pennsylvania to be ‘certain,’ as it must be, that . . . personnel do not advance the religious mission of the church-related schools in which they serve. Public Funds for Public Schools v. Marburger, 358 F. Supp. 29, 40-41, aff’d, 417 U. S. 961.” 421 U. S., at 370.

In Roemer v. Maryland Public Works Board, 426 U. S. 736 (1976), the Court sustained state programs of aid to religiously affiliated institutions of higher learning. The State allowed the grants to be used for any nonsectarian purpose. The Court upheld the grants on the ground that the institutions were not “‘pervasively sectarian,”’ id., at 758-759, and therefore a system of supervision was unnecessary to ensure that the grants were not being used to effect a religious end. In so holding, the Court identified “what is crucial to a non-entangling aid program: the ability of the State to identify and subsidize separate secular functions carried out at the school, without on-the-site inspections being necessary to prevent diversion of the funds to sectarian purposes.” Id., at 765. Similarly, in Tilton v. Richardson, 403 U. S. 672 (1971), the Court upheld one-time grants to sectarian institutions because ongoing supervision was not required. See also Hunt v. McNair, 413 U. S. 734 (1973).

As the Court of Appeals recognized, the elementary and secondary schools here are far different from the colleges at issue in Roemer, Hunt, and Tilton. 739 F. 2d; at 68-70. Unlike the colleges, which were found not to be “pervasively sectarian,” many of the schools involved in this case are the same sectarian schools which had “ ‘as a substantial purpose the inculcation of religious values’ ” in Committee for Public Education & Religious Liberty v. Nyquist, 413 U. S. 756, 768 (1973), quoting Committee for Public Education & Religious Liberty v. Nyquist, 350 F. Supp. 655, 663 (SDNY 1972). Moreover, our holding in Meek invalidating instructional services much like those at issue in this case rested *412on the ground that the publicly funded teachers were “performing important educational services in schools in which education is an integral part of the dominant sectarian mission and in which an atmosphere dedicated to the advancement of religious belief is constantly maintained.” Meek, supra, at 371. The court below found that the schools involved in this case were “well within this characterization.” 739 F. 2d, at 70.8 Unlike the schools in Roemer, many of the schools here receive funds and report back to their affiliated church, require attendance at church religious exercises, begin the schoolday or class period with prayer, and grant preference in admission to members of the sponsoring denominations. 739 F. 2d, at 70. In addition, the Catholic schools at issue here, which constitute the vast majority of the aided schools, are under the general supervision and control of the local parish. Ibid.

The critical elements of the entanglement proscribed in Lemon and Meek are thus present in this case. First, as noted above, the aid is provided in a pervasively sectarian environment. Second, because assistance is provided in the form of teachers, ongoing inspection is required to ensure the absence of a religious message. Compare Lemon, supra, at 619, with Tilton, supra, at 688, and Roemer, supra, at 765. In short, the scope and duration of New York City’s Title I *413program would require a permanent and pervasive state presence in the sectarian schools receiving aid.

This pervasive monitoring by public authorities in the sectarian schools infringes precisely those Establishment Clause values at the root of the prohibition of excessive entanglement. Agents of the city must visit and inspect the religious school regularly, alert for the subtle or overt presence of religious matter in Title I classes. Cf. Lemon v. Kurtzman, 403 U. S., at 619 (“What would appear to some to be essential to good citizenship might well for others border on or constitute instruction in religion”). In addition, the religious school must obey these same agents when they make determinations as to what is and what is not a “religious symbol” and thus off limits in a Title I classroom. In short, the religious school, which has as a primary purpose the advancement and preservation of a particular religion must endure the ongoing presence of state personnel whose primary purpose is to monitor teachers and students in an attempt to guard against the infiltration of religious thought.

The administrative cooperation that is required to maintain the educational program at issue here entangles church and state in still another way that infringes interests at the heart of the Establishment Clause. Administrative personnel of the public and parochial school systems must work together in resolving matters related to schedules, classroom assignments, problems that arise in the implementation of the program, requests for additional services, and the dissemination of information regarding the program. Furthermore, the program necessitates “frequent contacts between the regular and the remedial teachers (or other professionals), in which each side reports on individual student needs, problems encountered, and results achieved.” 739 F. 2d, at 65.

We have long recognized that underlying the Establishment Clause is “the objective ... to prevent, as far as possible, the intrusion of either [church or state] into the precincts of the other.” Lemon v. Kurtzman, supra, at 614. *414See also McCollum v. Board of Education, 333 U. S., at 212. Although “[separation in this context cannot mean absence of all contact,” Walz v. Tax Comm’n, 397 U. S. 664, 676 (1970), the detailed monitoring and close administrative contact required to maintain New York City’s Title I program can only produce “a kind of continuing day-to-day relationship which the policy of neutrality seeks to minimize.” Id., at 674. The numerous judgments that must be made by agents of the city concern matters that may be subtle and controversial, yet may be of deep religious significance to the controlling denominations. As government agents must make these judgments, the dangers of political divisiveness along religious lines increase. At the same time, “[t]he picture of state inspectors prowling the halls of parochial schools and auditing classroom instruction surely raises more than an imagined specter of governmental ‘secularization of a creed.’” Lemon v. Kurtzman, supra, at 650 (opinion of Brennan, J.).

Ill

Despite the well-intentioned efforts taken by the City of New York, the program remains constitutionally flawed owing to the nature of the aid, to the institution receiving the aid, and to the constitutional principles that they implicate— that neither the State nor Federal Government shall promote or hinder a particular faith or faith generally through the advancement of benefits or through the excessive entanglement of church and state in the administration of those benefits.

Affirmed.

[For dissenting opinion of Justice White, see ante, p. 400.]

concurring.

I concur in the Court’s opinions and judgments today in this case and in School District of Grand Rapids v. Ball, ante, p. 373, holding that the aid to parochial schools involved in those cases violates the Establishment Clause of the First *415Amendment. I write to emphasize additional reasons why precedents of this Court require us to invalidate these two educational programs that concededly have “done so much good and little, if any, detectable harm.” 739 F. 2d 48, 72 (CA2 1984). The Court has previously recognized the important role of parochial schools:

“‘Parochial schools, quite apart from their sectarian purpose, have provided an educational alternative for millions of young Americans; they often afford wholesome competition with our public schools; and in some States they relieve substantially the tax burden incident to the operation of public schools.’” Mueller v. Allen, 463 U. S. 388, 401-402 (1983) (quoting Wolman v. Walter, 433 U. S. 229, 262 (1977) (Powell, J., concurring in part, concurring in judgment in part, and dissenting in part)).

“The State has, moreover, a legitimate interest in facilitating education of the highest quality for all children within its boundaries, whatever school their parents have chosen for them.” 433 U. S., at 262. Regrettably, however, the Title I and Grand Rapids programs do not survive the scrutiny required by our Establishment Clause cases.

I agree with the Court that in this case the Establishment Clause is violated because there is too great a risk of government entanglement in the administration of the religious schools; the same is true in Ball, ante, p. 373. As beneficial as the Title I program appears to be in accomplishing its secular goal of supplementing the education of deprived children, its elaborate structure, the participation of public school teachers, and the government surveillance required to ensure that public funds are used for secular purposes inevitably present a serious risk of excessive entanglement. Our cases have noted that “‘[t]he State must be certain, given the Religion Clauses, that subsidized teachers do not inculcate religion.’” Meek v. Pittenger, 421 U. S. 349, 371 (1975) (emphasis added) (quoting Lemon v. Kurtzman, 403 *416U. S. 602, 619 (1971)). This is true whether the subsidized teachers are religious school teachers, as in Lemon, or public school teachers teaching secular subjects to parochial school children at the parochial schools. Judge Friendly, writing for the unanimous Court of Appeals, agreed with this assessment of our cases. He correctly observed that the structure of the Title I program required the active and extensive surveillance that the City has provided, and, “under Meek, this very surveillance constitutes excessive entanglement even if it has succeeded in preventing the fostering of religion.” 739 F. 2d, at 66.

This risk of entanglement is compounded by the additional risk of political divisiveness stemming from the aid to religion at issue here. I do not suggest that at this point in our history the Title I program or similar parochial aid plans could result in the establishment of a state religion. There likewise is small chance that these programs would result in significant religious or denominational control over our democratic processes. See Wolman v. Walter, supra, at 263 (Powell, J., concurring in part, concurring in judgment in part, and dissenting in part). Nonetheless, there remains a considerable risk of continuing political strife over the propriety of direct aid to religious schools and the proper allocation of limited governmental resources. As this Court has repeatedly recognized, there is a likelihood whenever direct governmental aid is extended to some groups that there will be competition and strife among them and others to gain, maintain, or increase the financial support of government. E. g., Committee for Public Education & Religious Liberty v. Nyquist, 413 U. S. 756, 796-797 (1973); Lemon v. Kurtzman, supra, at 623. In States such as New York that have large and varied sectarian populations, one can be assured that politics will enter into any state decision to aid parochial schools. Public schools, as well as private schools, are under increasing financial pressure to meet real and perceived needs. Thus, any proposal to extend direct governmental *417aid to parochial schools alone is likely to spark political disagreement from taxpayers who support the public schools, as well as from nonrecipient sectarian groups, who may fear that needed funds are being diverted from them. In short, aid to parochial schools of the sort at issue here potentially leads to “that kind and degree of government involvement in religious life that, as history teaches us, is apt to lead to strife and frequently strain a political system to the breaking point.” Walz v. Tax Comm’n, 397 U. S. 664, 694 (1970) (opinion of Harlan, J.). Although the Court’s opinion does not discuss it at length, see ante, at 413, the potential for such divisiveness is a strong additional reason for holding that the Title I and Grand Rapids programs are invalid on entanglement grounds.

The Title I program at issue in this case also would be invalid under the “effects” prong of the test adopted in Lemon v. Kurtzman, supra. * As has been discussed thoroughly in Ball, ante, at 392-397, with respect to the Grand Rapids programs, the type of aid provided in New York by the Title I program amounts to a state subsidy of the parochial schools by relieving those schools of the duty to provide the remedial and supplemental education their children require. This is not the type of “indirect and incidental effect beneficial to [the] religious institutions” that we suggested in Nyquist would survive Establishment Clause scrutiny. 413 U. S., at 775. Rather, by directly assuming part of the parochial schools’ education function, the effect of the Title I aid is “inevitably ... to subsidize and advance the religious mission of [the] sectarian schools,” id., at 779-780, even though the program provides that only secular subjects will *418be taught. As in Meek v. Pittenger, 421 U. S. 349 (1975), the secular education these schools provide goes “‘hand in hand’” with the religious mission that is the reason for the schools’ existence. 421 U. S., at 366 (quoting Lemon v. Kurtzman, 403 U. S., at 657 (opinion of Brennan, J.)). Because of the predominantly religious nature of the schools, the substantial aid provided by the Title I program “inescapably results in the direct and substantial advancement of religious activity.” Meek v. Pittenger, supra, at 366.

I recognize the difficult dilemma in which governments are placed by the interaction of the “effects” and entanglement prongs of the Lemon test. Our decisions require governments extending aid to parochial schools to tread an extremely narrow line between being certain that the “principal or primary effect” of the aid is not to advance religion, Lemon v. Kurtzman, supra, at 612, and avoiding excessive entanglement. Nonetheless, the Court has never foreclosed the possibility that some types of aid to parochial schools could be valid under the Establishment Clause. Mueller v. Allen, 463 U. S., at 393. Our cases have upheld evenhanded secular assistance to both parochial and public school children in some areas. E. g., ibid, (tax deductions for educational expenses); Board of Education v. Allen, 392 U. S. 236 (1968) (provision of secular textbooks); Everson v. Board of Education, 330 U. S. 1 (1947) (reimbursements for bus fare to school). I do not read the Court’s opinion as precluding these types of indirect aid to parochial schools. In the cases cited, the assistance programs made funds available equally to public and nonpublic schools without entanglement. The constitutional defect in the Title I program, as indicated above, is that it provides a direct financial subsidy to be administered in significant part by public school teachers within parochial schools — resulting in both the advancement of religion and forbidden entanglement. If, for example, Congress could fashion a program of evenhanded financial assistance to both public and private schools that could *419be administered, without governmental supervision in the private schools, so as to prevent the diversion of the aid from secular purposes, we would be presented with a different question.

I join the opinions and judgments of the Court;

dissenting.

Under the guise of protecting Americans from the evils of an Established Church such as those of the 18th century and earlier times, today’s decision will deny countless schoolchildren desperately needed remedial teaching services funded under Title I. The program at issue covers remedial reading, reading skills, remedial mathematics, English as a second language, and assistance for children needing special help in the learning process. The “remedial reading” portion of this program, for example, reaches children who suffer from dyslexia, a disease known to be difficult to diagnose and treat. Many of these children now will not receive the special training they need, simply because their parents desire that they attend religiously affiliated schools.

What is disconcerting about the result reached today is that, in the face of the human cost entailed by this decision, the Court does not even attempt to identify any threat to religious liberty posed by the operation of Title I. I share Justice White’s concern that the Court’s obsession with the criteria identified in Lemon v. Kurtzman, 403 U. S. 602 (1971), has led to results that are “contrary to the long-range interests of the country,” ante, at 400. As I wrote in Wallace v. Jaffree, 472 U. S. 38, 89 (1985) (dissenting opinion), “our responsibility is not to apply tidy formulas by rote; our duty is to determine whether the statute or practice at issue is a step toward establishing a state religion.” Federal programs designed to prevent a generation of children from growing up without being able to read effectively are not remotely steps in that direction. It borders on paranoia to perceive the Archbishop of Canterbury or the Bishop of *420Rome lurking behind programs that are just as vital to the Nation’s schoolchildren as textbooks, see generally Board of Education v. Allen, 392 U. S. 236 (1968), transportation to and from school, see generally Everson v. Board of Education, 330 U. S. 1 (1947), and school nursing services.

On the merits of this case, I dissent for the reasons stated in my separate opinion in Meek v. Pittenger, 421 U. S. 349 (1975). We have frequently recognized that some interaction between church and state is unavoidable, and that an attempt to eliminate all contact between the two would be both futile and undesirable. Justice Douglas, writing for the Court in Zorach v. Clauson, 343 U. S. 306, 312 (1952), stated:

“The First Amendment . . . does not say that in every and all respects there shall be a separation of Church and State. . . . Otherwise the state and religion would be aliens to each other — hostile, suspicious, and even unfriendly.”

The Court today fails to demonstrate how the interaction occasioned by the program at issue presents any threat to the values underlying the Establishment Clause.

I cannot join in striking down a program that, in the words of the Court of Appeals, “has done so much good and little, if any, detectable harm.” 739 F. 2d 48, 72 (CA2 1984). The notion that denying these services to students in religious schools is a neutral act to protect us from an Established Church has no support in logic, experience, or history. Rather than showing the neutrality the Court boasts of, it exhibits nothing less than hostility toward religion and the children who attend church-sponsored schools.

dissenting.

I dissent for the reasons stated in my dissenting opinion in Wallace v. Jaffree, 472 U. S. 38, 91 (1985). In this case the Court takes advantage of the “Catch-22” paradox of its own creation, see Wallace, supra, at 109-110 (Rehnquist, J., *421dissenting), whereby aid must be supervised to ensure no entanglement but the supervision itself is held to cause an entanglement. The Court today strikes down nondiscriminatory nonsectarian aid to educationally deprived children from low-income families. The Establishment Clause does not prohibit such sorely needed assistance; we have indeed traveled far afield from the concerns which prompted the adoption of the First Amendment when we rely on gossamer abstractions to invalidate a law which obviously meets an entirely secular need. I would reverse.

with whom Justice Rehnquist joins as to Parts II and III, dissenting.

Today the Court affirms the holding of the Court of Appeals that public school teachers can offer remedial instruction to disadvantaged students who attend religious schools “only if such instruction . . . [is] afforded at a neutral site off the premises of the religious school.” 739 F. 2d 48, 64 (CA2 1984). This holding rests on the theory, enunciated in Part V of the Court’s opinion in Meek v. Pittenger, 421 U. S. 349, 367-373 (1975), that public school teachers who set foot on parochial school premises are likely to bring religion into their classes, and that the supervision necessary to prevent religious teaching would unduly entangle church and state. Even if this theory were valid in the abstract, it cannot validly be applied to New York City’s 19-year-old Title I program. The Court greatly exaggerates the degree of supervision necessary to prevent public school teachers from inculcating religion, and thereby demonstrates the flaws of a test that condemns benign cooperation between church and state. I would uphold Congress’ efforts to afford remedial instruction to disadvantaged schoolchildren in both public and parochial schools.

I

As in Wallace v. Jaffree, 472 U. S. 38 (1985), and Thornton v. Caldor, Inc., 472 U. S. 703 (1985), the Court in this litigation adheres to the three-part Establishment Clause *422test enunciated in Lemon v. Kurtzman, 403 U. S. 602, 612-613 (1971). To survive the Lemon test, a statute must have both a secular legislative purpose and a principal or primary effect that neither advances nor inhibits religion. Under Lemon and its progeny, direct state aid to parochial schools that has the purpose or effect of furthering the religious mission of the schools is unconstitutional. I agree with that principle. According to the Court, however, the New York City Title I program is defective not because of any improper purpose or effect, but rather because it fails the third part of the Lemon test: the Title I program allegedly fosters excessive government entanglement with religion. I disagree with the Court’s analysis of entanglement, and I question the utility of entanglement as a separate Establishment Clause standard in most cases. Before discussing entanglement, however, it is worthwhile to explore the purpose and effect of the New York City Title I program in greater depth than does the majority opinion.

The purpose of Title I is to provide special educational assistance to disadvantaged children who would not otherwise receive it. Congress recognized that poor academic performance by disadvantaged children is part of the cycle of poverty. S. Rep. No. 146, 89th Cong., 1st Sess., 4 (1965). Congress sought to break the cycle by providing classes in remedial reading, mathematics, and English to disadvantaged children in parochial as well as public schools, for public schools enjoy no monopoly on education in low-income areas. Wheeler v. Barrera, 417 U. S. 402, 405-406 (1974). See 20 U. S. C. §§ 2740(a), 3806(a). Congress permitted remedial instruction by public school teachers on parochial school premises only if such instruction is “not normally provided by the nonpublic school” and would “contribute particularly to meeting the special educational needs of educationally deprived children.” S. Rep. No. 146, supra, at 12. See 34 CFR §200.73 (1984) (Department of Education regulations implementing Title I and precluding instruction on parochial *423school premises except where necessary and where such instruction is not normally provided by the school).

After reviewing the text of the statute and its legislative history, the District Court concluded that Title I serves a secular purpose of aiding needy children regardless of where they attend school. App. to Juris. Statement in No. 84-238, p. 56a, incorporating findings of the District Court in National Coalition for Public Education and Religious Liberty v. Harris, 489 F. Supp. 1248, 1258 (SDNY 1980) (PEARL). The Court of Appeals did not dispute this finding, and no party in this Court contends that the purpose of the statute or of the New York City Title I program is to advance or endorse religion. Indeed, the record demonstrates that New York City public school teachers offer Title I classes on the premises of parochial schools solely because alternative means to reach the disadvantaged parochial school students — such as instruction for parochial school students at the nearest public school, either after or during regular school hours — were unsuccessful. PEARL, supra, at 1255. As the Court of Appeals acknowledged, New York City “could reasonably have regarded [Title I instruction on parochial school premises] as the most effective way to carry out the purposes of the Act.” 739 F. 2d, at 49. Whether one looks to the face of the statute or to its implementation, the Title I program is undeniably animated by a legitimate secular purpose.

The Court’s discussion of the effect of the New York City Title I program is even more perfunctory than its analysis of the program’s purpose. The Court’s opinion today in School District of Grand Rapids v. Ball, ante, p. 373, which strikes down a Grand Rapids scheme that the Court asserts is very similar to the New York City program, identifies three ways in which public instruction on parochial school premises may have the impermissible effect of advancing religion. First, “state-paid instructors, influenced by the pervasively sectarian nature of the religious schools in which they work, may *424subtly or overtly indoctrinate the students in particular religious tenets at public expense.” Second, “state-provided instruction in the religious school buildings threatens to convey a message of state support for religion to students and to the general public.” Third, “the programs in effect subsidize the religious functions of the parochial schools by taking over a substantial portion of their responsibility for teaching secular subjects.” Ante, at 397. While addressing the effect of the Grand Rapids program at such length, the Court overlooks the effect of Title I in New York City.

One need not delve too deeply in the record to understand why the Court does not belabor the effect of the Title I program. The abstract theories explaining why on-premises instruction might possibly advance religion dissolve in the face of experience in New York City. As the District Court found in 1980:

“New York City has been providing Title I services in nonpublic schools for fourteen years. The evidence presented in this action includes: extensive background information on Title I; an in-depth description of New York City’s program; a detailed review of Title I rules and regulations and the ways in which they are enforced; and the testimony and affidavits of federal officials, state officers, school administrators, Title I teachers and supervisors, and parents of children receiving Title I services. The evidence establishes that the result feared in other cases has not materialized in the City’s Title I program. The presumption — that the ‘religious mission’ will be advanced by providing educational services on parochial school premises — is not supported by the facts of this case.” PEARL, supra, at 1265.

Indeed, in 19 years there has never been a single incident in which a Title I instructor “subtly or overtly” attempted to “indoctrinate the students in particular religious tenets at public expense.” Grand Rapids, ante, at 397.

*425Common sense suggests a plausible explanation for this unblemished record. New York City’s public Title I instructors are professional educators who can and do follow instructions not to inculcate religion in their classes. They are unlikely to be influenced by the sectarian nature of the parochial schools where they teach, not only because they are carefully supervised by public officials, but also because the vast majority of them visit several different schools each week and are not of the same religion as their parochial students.* In light of the ample record, an objective observer of the implementation of the Title I program in New York City would hardly view it as endorsing the tenets of the participating parochial schools. To the contrary, the actual and perceived effect of the program is precisely the effect intended by Congress: impoverished schoolchildren are being helped to overcome learning deficits, improving their test scores, and receiving a significant boost in their struggle to obtain both a thorough education and the opportunities that flow from it.

The only type of impermissible effect that arguably could carry over from the Grand, Rapids decision to this litigation, then, is the effect of subsidizing “the religious functions of the parochial schools by taking over a substantial portion of their responsibility for teaching secular subjects.” Ibid. That effect is tenuous, however, in light of the statutory directive that Title I funds may be used only to provide services that otherwise would not be available to the participating students. 20 U. S. C. § 3807(b). The Secretary of Education has vigorously enforced the requirement that Title I funds supplement rather than supplant the services of local education agencies. See Bennett v. Kentucky Dept. of Ed., 470 U. S. 656 (1985); Bennett v. New Jersey, 470 U. S. 632 (1985).

*426Even if we were to assume that Title I remedial classes in New York City may have duplicated to some extent instruction parochial schools would have offered in the absence of Title I, the Court’s delineation of this third type of effect proscribed by the Establishment Clause would be seriously flawed. Our Establishment Clause decisions have not barred remedial assistance to parochial school children, but rather remedial assistance on the premises of the parochial school. Under Wolman v. Walter, 433 U. S. 229, 244-248 (1977), the New York City classes prohibited by the Court today would have survived Establishment Clause scrutiny if they had been offered in a neutral setting off the property of the private school. Yet it is difficult to understand why a remedial reading class offered on parochial school premises is any more likely to supplant the secular course offerings of the parochial school than the same class offered in a portable classroom next door to the school. Unless Wolman was wrongly decided, the defect in the Title I program cannot lie in the risk that it will supplant secular course offerings.

I — I I — I

Recognizing the weakness of any claim of an improper purpose or effect, the Court today relies entirely on the entanglement prong of Lemon to invalidate the New York City Title I program. The Court holds that the occasional presence of peripatetic public school teachers on parochial school grounds threatens undue entanglement of church and state because (1) the remedial instruction is afforded in a pervasively sectarian environment; (2) ongoing supervision is required to assure that the public school teachers do not attempt to inculcate religion; (3) the administrative personnel of the parochial and public school systems must work together in resolving administrative and scheduling problems'; and (4) the instruction is likely to result in political divisiveness over the propriety of direct aid. Ante, at 412-414; ante, at 415-416 (concurring opinion of Powell, J.).

*427This analysis of entanglement, I acknowledge, finds support in some of this Court’s precedents. In Meek v. Pittenger, 421 U. S., at 369, the Court asserted that it could not rely “on the good faith and professionalism of the secular teachers and counselors functioning in church-related schools to ensure that a strictly nonideological posture is maintained.” Because “a teacher remains a teacher,” the Court stated, there remains a risk that teachers will intertwine religious doctrine with secular instruction. The continuing state surveillance necessary to prevent this from occurring would produce undue entanglement of church and state. Id., at 370-372. The Court’s opinion in Meek further asserted that public instruction on parochial school premises creates a serious risk of divisive political conflict over the issue of aid to religion. Ibid. Meek’s analysis of entanglement was reaffirmed in Wolman two Terms later.

I would accord these decisions the appropriate deference commanded by the doctrine of stare decisis if I could discern logical support for their analysis. But experience has demonstrated that the analysis in Part V of the Meek opinion is flawed. At the time Meek was decided, thoughtful dissents pointed out the absence of any record support for the notion that public school teachers would attempt to. inculcate religion simply because they temporarily occupied a parochial school classroom, or that such instruction would produce political divisiveness. Id., at 386 (opinion of Burger, C. J.); id., at 387 (opinion of Rehnquist, J.). Experience has given greater force to the arguments of the dissenting opinions in Meek. It is not intuitively obvious that a dedicated public school teacher will tend to disobey instructions and commence proselytizing students at public expense merely because the classroom is within a parochial school. Meek is correct in asserting that a teacher of remedial reading “remains a teacher,” but surely it is significant that the teacher involved is a professional, full-time public school employee who is unaccustomed to bringing religion into the classroom. *428Given that not a single incident of religious indoctrination has been identified as occurring in the thousands of classes offered in Grand Rapids and New York City over the past two decades, it is time to acknowledge that the risk identified in Meek was greatly exaggerated.

Just as the risk that public school teachers in parochial classrooms will inculcate religion has been exaggerated, so has the degree of supervision required to manage that risk. In this respect the New York City Title I progam is instructive. What supervision has been necessary in New York City to enable public school teachers to help disadvantaged children for 19 years without once proselytizing? Public officials have prepared careful instructions warning public school teachers of their exclusively secular mission, and have required Title I teachers to study and observe them. App. 50-51. Under the rules, Title I teachers are not accountable to parochial or private school officials; they have sole responsibility for selecting the students who participate in their class, must administer their own tests for determining eligibility, cannot engage in team teaching or cooperative activities with parochial school teachers, must make sure that all materials and equipment they use are not otherwise used by the parochial school, and must not participate in religious activities in the schools or introduce any religious matter into their teaching. To ensure compliance with the rules, a field supervisor and a program coordinator, who are full-time public school employees, make unannounced visits to each teacher’s classroom at least once a month. Id., at 53.

The Court concludes that this degree of supervision of public school employees by other public school employees constitutes excessive entanglement of church and state. I cannot agree. The supervision that occurs in New York City’s Title I program does not differ significantly from the supervision any public school teacher receives, regardless of the location of the classroom. Justice Powell suggests that the required supervision is extensive because the State must be *429certain that public school teachers do not inculcate religion. Ante, at 415. That reasoning would require us to close our public schools, for there is always some chance that a public school teacher will bring religion into the classroom, regardless of its location. See Wallace v. Jaffree, 472 U. S., at 44-45, n. 23. Even if I remained confident of the usefulness of entanglement as an Establishment Clause test, I would conclude that New York City’s efforts to prevent religious indoctrination in Title I classes have been adequate and have not caused excessive institutional entanglement of church and state.

The Court’s rebanee on the potential for political divisiveness as evidence of undue entanglement is also unpersuasive. There is little record support for the proposition that New York City’s admirable Title I program has ignited any controversy other than this litigation. In Mueller v. Allen, 463 U. S. 388, 403-404, n. 11 (1983), the Court cautioned that the “elusive inquiry” into political divisiveness should be confined to a narrow category of parochial aid cases. The concurring opinion in Lynch v. Donnelly, 465 U. S. 668, 687 (1984), went further, suggesting that Establishment Clause analysis should focus solely on the character of the government activity that might cause political divisiveness, and that “the entanglement prong of the Lemon test is properly limited to institutional entanglement.”

I adhere to the doubts about the entanglement test that were expressed in Lynch. It is curious indeed to base our interpretation of the Constitution on speculation as to the likelihood of a phenomenon which the parties may create merely by prosecuting a lawsuit. My reservations about the entanglement test, however, have come to encompass its institutional aspects as well. As Justice Rehnquist has pointed out, many of the inconsistencies in our Establishment Clause decisions can be ascribed to our insistence that parochial aid programs with a valid purpose and effect may still be invalid by virtue of undue entanglement. Wallace v. *430Jaffree, supra, at 109-110. For example, we permit a State to pay for bus transportation to a parochial school, Everson v. Board of Education, 330 U. S. 1 (1947), but preclude States from providing buses for parochial school field trips, on the theory such trips involve excessive state supervision of the parochial officials who lead them. Wolman, 433 U. S., at 254. To a great extent, the anomalous results in our Establishment Clause cases are “attributable to [the] ‘entanglement’ prong.” Choper, The Religion Clauses of the First Amendment: Reconciling the Conflict, 41 U. Pitt. L. Rev. 673, 681 (1980).

Pervasive institutional involvement of church and state may remain relevant in deciding the effect of a statute which is alleged to violate the Establishment Clause, Walz v. Tax Comm’n, 397 U. S. 664 (1970), but state efforts to ensure that public resources are used only for nonsectarian ends should not in themselves serve to invalidate an otherwise valid statute. The State requires sectarian organizations to cooperate on a whole range of matters without thereby advancing religion or giving the impression that the government endorses religion. Wallace v. Jaffree, supra, at 110 (dissenting opinion of Rehnquist, J.) (noting that state educational agencies impose myriad curriculum, attendance, certification, fire, and safety regulations on sectarian schools). If a statute lacks a purpose or effect of advancing or endorsing religion, I would not invalidate it merely because it requires some ongoing cooperation between church and state or some state supervision to ensure that state funds do not advance religion.

Ill

Today’s ruling does not spell the end of the Title I program of remedial education for disadvantaged children. Children attending public schools may still obtain the benefits of the program. Impoverished children who attend parochial schools may also continue to benefit from Title I programs offered off the premises of their schools — possibly in portable *431classrooms just over the edge of school property. The only disadvantaged children who lose under the Court’s holding are those in cities where it is not economically and logistically feasible to provide public facilities for remedial education adjacent to the parochial school. But this subset is significant, for it includes more than 20,000 New York City schoolchildren and uncounted others elsewhere in the country.

For these children, the Court’s decision is tragic. The Court deprives them of a program that offers a meaningful chance at success in life, and it does so on the untenable theory that public school teachers (most of whom are of different faiths than their students) are likely to start teaching religion merely because they have walked across the threshold of a parochial school. I reject this theory and the analysis in Meek v. Pittenger on which it is based. I cannot close my eyes to the fact that, over almost two decades, New York City’s public school teachers have helped thousands of impoverished parochial school children to overcome educational disadvantages without once attempting to inculcate religion. Their praiseworthy efforts have not eroded and do not threaten the religious liberty assured by the Establishment Clause. The contrary judgment of the Court of Appeals should be reversed.

I respectfully dissent.