Plaintiffs in these three consolidated cases are Ohio prisoners who contend, among other claims, that various Ohio corrections officials have violated the Religious Land Use and Institutionalized Persons Act (RLUIPA), 42 U.S.C. § 2000cc-2000cc-5. The defendant officials filed motions to dismiss the RLUIPA claims, challenging the constitutionality of 42 U.S.C. § 2000cc-1, the section of the Act that applies to institutionalized persons. Their motions were denied by the district court. In this interlocutory appeal, defendants contend that RLUIPA (1) exceeds Congress's powers under both the Spending and Commerce Clauses, (2) violates states' rights under the Tenth Amendment, and (3) improperly advances religion in violation of the Establishment Clause of the First Amendment. For the reasons set forth below, we agree that the portion of RLUIPA that applies to institutionalized persons — specifically, 42 U.S.C. § 2000cc-1 — violates the Establishment Clause. We therefore REVERSE the district court's denial of defendants' motions to dismiss and REMAND the case for further proceedings consistent with this opinion.

Because this appeal involves a facial challenge to RLUIPA, the facts of the individual cases are not particularly relevant. The prisoners in all three cases generally allege that officials with the Ohio Department of Rehabilitation and Corrections (ODRC) violated RLUIPA by refusing to accommodate the prisoners' religious beliefs and practices. Defendants, on the other hand, contend that RLUIPA has allowed inmate gangs to claim "`religious' status in order to insulate their illicit activities from scrutiny."

What is relevant to this case is the history and substance of RLUIPA. In 1990, the Supreme Court held that the United States Constitution does not require that government have a compelling state interest in order to enact a law of general applicability that incidentally burdens the exercise of religion. Employment Div., Dept. of Human Res. v. Smith, 494 U.S. 872, 110 S. Ct. 1595, 108 L. Ed. 2d 876 (1990). Congress responded in 1993 by enacting the Religious Freedom Restoration Act (RFRA), 42 U.S.C. §§ 2000bb-2000bb-4. RFRA required that any governmental attempt to "substantially burden" the exercise of religion must be the least restrictive means of furthering a compelling state interest. 42 U.S.C. § 2000bb 1(b). The Supreme Court held RFRA unconstitutional insofar as it applied to states and localities because the statute exceeded Congress's powers under the Fourteenth Amendment. City of Boerne v. Flores, 521 U.S. 507, 117 S. Ct. 2157, 138 L. Ed. 2d 624 (1997).

Congress reacted to Boerne by passing RLUIPA in 2000. RLUIPA has the same substantive standard as RFRA. It provides, in relevant part, that "[n]o government shall impose a substantial burden on the religious exercise of a person residing in or confined to an institution" unless the burden "is in furtherance of a compelling governmental interest" and "is the least restrictive means" of furthering that interest. 42 U.S.C. § 2000cc-1(a). The Act defines "religious exercise" as "any exercise of religion, whether or not compelled by, or central to, a system of religious belief." 42 U.S.C. § 2000cc-5(7)(A). RLUIPA's requirement of strict scrutiny stands in sharp contrast to the Supreme Court's previous decisions, which have held that the courts should apply a rational-relationship review to restrictions upon inmates' fundamental rights. See Turner v. Safley, 482 U.S. 78, 107 S. Ct. 2254, 96 L. Ed. 2d 64 (1987) (applying the rational-relationship test to prison rules regulating prisoner correspondence and marriage); O'Lone v. Estate of Shabazz, 482 U.S. 342, 107 S. Ct. 2400, 96 L. Ed. 2d 282 (1987) (applying the rational-relationship test to prison rules regulating prisoners' religious exercise).

Congress enacted RLUIPA pursuant to its powers under the Spending Clause, U.S. Const. art. I, § 8, cl. 1, and the Commerce Clause, U.S. Const. art. I, § 8, cl. 3. RLUIPA applies where "the substantial burden [on religious exercise] is imposed in a program or activity that receives Federal financial assistance." 42 U.S.C. § 2000cc-1(b)(1). The Act is also applicable where "the substantial burden affects, or removal of that substantial burden would affect, commerce with foreign nations, among the several States, or with Indian tribes." 42 U.S.C. § 2000cc-1(b)(2).

RLUIPA creates a private right of action. Any person may "assert a violation of this chapter as a claim or defense in a judicial proceeding" and may obtain "appropriate relief against a government." 42 U.S.C. § 2000cc-2(a). The United States may also seek injunctive or declaratory relief to enforce the statute. 42 U.S.C. § 2000cc-2(f).

RLUIPA's congressional sponsors specifically noted that they expected federal courts to respect the decisions of prison officials as to what restrictions on the exercise of religion are necessary in the prison context. A joint statement to the Senate expressed the sponsors' belief that federal courts would "continue the tradition of giving due deference to the experience and expertise of prison and jail administrators in establishing necessary regulations and procedures to maintain good order, security, and discipline, consistent with considerations of cost and limited resources." Statements of Senators Hatch and Kennedy, 146 Cong. Rec. S7774-01, S7775 (2000).

The plaintiff prisoners brought suit against various Ohio corrections officials based upon claims arising under the First and Fourteenth Amendments. RLUIPA went into effect after suit was filed, causing plaintiffs to amend their complaints to include claims under the Act. Defendants then filed motions to dismiss the RLUIPA claims, arguing that the Act was unconstitutional. (All references to RLUIPA are to 42 U.S.C. § 2000cc-1 only, the portion of the Act that applies to institutionalized persons.) The United States intervened to defend the constitutionality of the Act. All three cases were consolidated in order to adjudicate the motions to dismiss at the same time.

On August 27, 2001, the magistrate judge filed a Report and Recommendation, which concluded that the statute was constitutional and recommended that the district court deny defendants' motions to dismiss. The district court entered an opinion and order adopting the Report and Recommendation on February 25, 2002. Approximately a year and a half later, on August 4, 2003, the court certified its February 25, 2002 order for immediate appeal pursuant to 28 U.S.C. § 1292(b). The parties then filed a joint petition for leave to appeal with this court on August 6, 2003, within the 10-day time limit imposed by the statute. We have previously granted the petition for leave to appeal.

The Supreme Court has not yet considered the constitutionality of RLUIPA. Justice Stevens, however, in his concurring opinion in Boerne, concluded that RLUIPA's predecessor, RFRA, violated the Establishment Clause:

In my opinion, the Religious Freedom Restoration Act of 1993 (RFRA) is a "law respecting an establishment of religion" that violates the First Amendment to the Constitution.... [T]he statute has provided [religious organizations] with a legal weapon that no atheist or agnostic can obtain. This governmental preference for religion, as opposed to irreligion, is forbidden by the First Amendment.

Boerne, 521 U.S. at 536-37, 117 S. Ct. 2157 (1997) (Stevens, J., concurring).

Two circuits, without reference to Justice Stevens's concurring opinion, have come to the opposite conclusion regarding the constitutionality of RLUIPA. See Charles v. Verhagen, 2003 WL 22455960 (7th Cir. Oct. 30, 2003); Mayweathers v. Newland, 314 F.3d 1062 (9th Cir.2002). Furthermore, five circuits, including the Seventh and Ninth, have concluded that the identical operative language in RFRA does not violate the Establishment Clause. See In re Young, 141 F.3d 854, 863 (8th Cir.1998); Mockaitis v. Harcleroad, 104 F.3d 1522, 1530 (9th Cir.1997); Sasnett v. Sullivan, 91 F.3d 1018, 1022 (7th Cir.1996), vacated on other grounds, 521 U.S. 1114, 117 S. Ct. 2502, 138 L. Ed. 2d 1007 (1997); EEOC v. Catholic Univ. of Am., 83 F.3d 455, 470 (D.C.Cir.1996); Flores v. City of Boerne, 73 F.3d 1352, 1364 (5th Cir.1996), rev'd on other grounds, 521 U.S. 507, 117 S. Ct. 2157, 138 L. Ed. 2d 624 (1997). Two district court opinions, in addition to the one below, have also concluded that RLUIPA is constitutional. See Johnson v. Martin, 223 F. Supp. 2d 820 (W.D.Mich.2002); Charles v. Verhagen, 220 F. Supp. 2d 955 (W.D.Wis.2002), aff'd, 2003 WL 22455960 (7th Cir. Oct. 30, 2003).

Against this apparent juggernaut of circuit and district court opinions stand two district court decisions that reach the opposite conclusion. One is Madison v. Riter, 240 F. Supp. 2d 566 (W.D.Va.2003) (Turk, J.), and the other is Kilaab Al Ghashiyah (Khan) v. Dep't of Corrections, 250 F. Supp. 2d 1016 (E.D.Wis.2003) (Adelman, J.), overruled by Charles v. Verhagen, 2003 WL 22455960 (7th Cir. Oct. 30, 2003). Both are remarkably well-worded and persuasive opinions that clearly set forth the history of RLUIPA, the analytical basis for concluding that RLUIPA violates the Establishment Clause, and the unpersuasive nature of the contrary opinions. Indeed, our own analysis can (and will) be considerably streamlined by repeated references to Madison and Ghashiyah. (Inexplicably, the Seventh Circuit in Charles makes no reference to either of these district court opinions.)

B. RLUIPA violates the Establishment Clause because it favors religious rights over other fundamental rights without any showing that religious rights are at any greater risk of deprivation

The Establishment Clause of the First Amendment to the U.S. Constitution states: "Congress shall make no law respecting an establishment of religion." Neutrality is the fundamental requirement of the Establishment Clause, which prohibits government from either endorsing a particular religion or promoting religion generally. Bd. of Educ. of Kiryas Joel Village Sch. Dist. v. Grumet, 512 U.S. 687, 703, 114 S. Ct. 2481, 129 L. Ed. 2d 546 (1994) ("[A] principle at the heart of the Establishment Clause [is] that government should not prefer one religion to another, or religion to irreligion."); see also Ghashiyah, 250 F.Supp.2d at 1021 (collecting cases that discuss the neutrality requirement).

In Lemon v. Kurtzman, 403 U.S. 602, 91 S. Ct. 2105, 29 L. Ed. 2d 745 (1971), the Supreme Court articulated a three-part test to determine whether a statute violates the Establishment Clause. A statute (1) "must have a secular legislative purpose," (2) "its principal or primary effect must be one that neither advances nor inhibits religion," and (3) it must not create "excessive government entanglement with religion." Id. at 612-13 (internal quotation marks and citations omitted). The Supreme Court suggested a modification to the Lemon test in Agostini v. Felton, 521 U.S. 203, 232-35, 117 S. Ct. 1997, 138 L. Ed. 2d 391 (1997), in the context of considering the constitutionality of government aid to parochial schools, where the Court proceeded to analyze entanglement under the effect prong rather than as a separate factor. Based upon our precedent of applying the Lemon test other than in aid-to-education cases, however, we will proceed with the traditional three-part Lemon analysis. See, e.g., Adland v. Russ, 307 F.3d 471, 479 (6th Cir.2002) (applying the Lemon test to decide that a Kentucky legislative resolution directing the state to move a Ten Commandments monument to a permanent site on the state capitol grounds violated the Establishment Clause).

"The purpose prong of the Lemon test asks whether government's actual purpose is to endorse or disapprove of religion." Edwards v. Aguillard, 482 U.S. 578, 585, 107 S. Ct. 2573, 96 L. Ed. 2d 510 (1987) (quoting Lynch v. Donnelly, 465 U.S. 668, 690, 104 S. Ct. 1355, 79 L. Ed. 2d 604 (1984) (O'Connor, J., concurring)). Lemon's requirement of a secular purpose "does not mean that the law's purpose must be unrelated to religion...." Corp. of the Presiding Bishop of the Church of Jesus Christ of Latter-Day Saints v. Amos, 483 U.S. 327, 335, 107 S. Ct. 2862, 97 L. Ed. 2d 273 (1987). Instead, the purpose prong "aims at preventing the relevant governmental decisionmaker ... from abandoning neutrality and acting with the intent of promoting a particular point of view in religious matters." Id.

In Amos, the Supreme Court considered whether Congress had violated the Establishment Clause by exempting religious organizations from Title VII's prohibition against religious discrimination in employment. Amos held 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." Id. In the present case, plaintiffs argue that RLUIPA has a virtually identical purpose: to alleviate significant interference by prison officials with the ability of prisoners to exercise their religious beliefs. But material differences exist between the application of RLUIPA in a prison setting and the Title VII exemption at issue in Amos.

One key difference is that the exemption in Amos was arguably necessary to avoid a violation of the Establishment Clause. Without the exemption, Title VII would have required courts to interfere with the internal workings of religious organizations, and fear of liability might have affected the way religious organizations carried out their missions. Id. at 336, 107 S. Ct. 2862; see also id. at 344, 107 S. Ct. 2862 (Brennan, J., concurring) ("A case-by-case analysis for all activities therefore would both produce excessive government entanglement with religion and create the danger of chilling religious activity."). Enacting RLUIPA, on the other hand, was not even arguably necessary to avoid a violation of the Establishment Clause. The Supreme Court had previously held that government interference with prisoners' fundamental rights is not subject to strict scrutiny, as RLUIPA requires, but only to a rational-relationship review. See Turner, 482 U.S. 78, 107 S. Ct. 2254, 96 L. Ed. 2d 64; O'Lone, 482 U.S. 342, 107 S. Ct. 2400, 96 L. Ed. 2d 282.

Another key difference between RLUIPA and the exemption in Amos is that RLUIPA sweeps much more broadly. As one commentator noted about RFRA, the predecessor to RLUIPA that has identical substantive provisions:

Comparing RFRA to Amos is like comparing apples to oranges. RFRA, unlike Amos, does not exempt religion from regulation for the purpose of avoiding an Establishment Clause violation. Rather, RFRA institutes a standard of review in every case which implicates religious conduct.... Amos did not involve a law that exempted religion from every law in the country. Rather, it permitted the exemption of religious employers from a particular requirement in prescribed circumstances. The law in Amos lacked RFRA's vast scope; therefore, Amos cannot dictate how RFRA fares under the Establishment Clause.

Marci A. Hamilton, The Religious Freedom Restoration Act is Unconstitutional, Period, 1 U. Pa. J. Const. L. 1, 13-14 (1998).

The broader scope of RLUIPA suggests that its actual purpose is not to accommodate religion by removing a particular obstacle to religious exercise, but "to advance religion in prisons relative to other constitutionally protected conduct." Ghashiyah, 250 F.Supp.2d at 1024. If that is indeed the true purpose of RLUIPA, then Congress has "abandoned neutrality and acted with the purpose of furthering religion," in violation of the Establishment Clause's fundamental command of governmental neutrality. Id. at 1025.

Resolution of the question of whether RLUIPA has the proper purpose of alleviating government interference with religious exercise or the prohibited purpose of advancing religion in prisons is not necessary to our ultimate decision regarding the Act's constitutionality. Even if the purpose of RLUIPA fits within the rule of Amos, RLUIPA is still unconstitutional because it has the primary effect of advancing religion. See Amos, 483 U.S. at 334-35, 107 S. Ct. 2862 ("At some point, accommodation may devolve into an unlawful fostering of religion ....") (internal quotation marks omitted).

"The effect prong [of the Lemon test] asks whether, irrespective of government's actual purpose, the practice under review in fact conveys a message of endorsement or disapproval." Lynch v. Donnelly, 465 U.S. 668, 690, 104 S. Ct. 1355, 79 L. Ed. 2d 604 (1984) (O'Connor, J., concurring). In evaluating this prong, the two most relevant factors are (1) whether a particular government action benefits both secular and religious entities, and (2) whether the action will induce religious exercise, rather than only protecting it. See Ghashiyah, 250 F.Supp.2d at 1025-26 (collecting authorities that have used these factors in Establishment Clause cases). Applying these factors to the present case demonstrates that RLUIPA has the effect of impermissibly advancing religion by giving greater protection to religious rights than to other constitutionally protected rights.

Prior to RLUIPA, restrictions imposed by prison officials upon inmates' fundamental rights were subject to a rational-relationship review, see Turner, 482 U.S. 78, 107 S. Ct. 2254, 96 L. Ed. 2d 64; O'Lone, 482 U.S. 342, 107 S. Ct. 2400, 96 L. Ed. 2d 282, which requires courts to consider: (1) whether there is a "valid, rational connection" between the prison regulation and a legitimate government interest; (2) whether inmates have alternative means of exercising the right in question; (3) the impact of a requested accommodation of the right upon guards and other inmates; and (4) the absence of alternatives to the regulation. Turner, 482 U.S. at 89-90, 107 S. Ct. 2254. The rational-relationship test has been applied to claimed violations of various fundamental rights, including the right to the free exercise of religion, O'Lone, 482 U.S. 342, 107 S. Ct. 2400, 96 L. Ed. 2d 282, the right to freedom of speech, Amatel v. Reno, 156 F.3d 192 (D.C.Cir.1998), the right to marry, Turner, 482 U.S. 78, 107 S. Ct. 2254, 96 L. Ed. 2d 64, the right to privacy, Oliver v. Scott, 276 F.3d 736 (5th Cir.2002), the right to meaningful access to the courts, Lewis v. Casey, 518 U.S. 343, 116 S. Ct. 2174, 135 L. Ed. 2d 606 (1996), and the right to be free from racial discrimination, Morrison v. Garraghty, 239 F.3d 648 (4th Cir.2001).

In contrast to the highly deferential rational-relationship test, RLUIPA requires courts to apply strict scrutiny to all substantial burdens upon the free exercise of religion. Madison eloquently explained the dramatic changes imposed by RLUIPA:

Under RLUIPA, prison regulations that substantially burden religious belief, including those that are generally applicable and facially neutral, are judged under a strict scrutiny standard, requiring prison officials, rather than the inmate, to bear the burden of proof that the regulation furthers a compelling penological interest and is the least restrictive means of satisfying this interest. 42 U.S.C. § 2000cc-1. As is well known from the history of constitutional law, the change that RLUIPA imposes is revolutionary, switching from a scheme of deference to one of presumptive unconstitutionality. See Smith, 494 U.S. at 888, 110 S. Ct. 1595. Instead of rational, the penological interest under RLUIPA must be of the highest order, see Wisconsin v. Yoder, 406 U.S. 205, 215, 92 S. Ct. 1526, 32 L. Ed. 2d 15 (1972); Jenkins v. Angelone, 948 F. Supp. 543, 546 (E.D.Va.1996); instead of focusing on the prison inmate's ability to find other avenues to exercise his belief, a court is required to focus on the prison administrator's choice among regulatory options, see 42 U.S.C. § 2000cc-1(a)(2); instead of placing the burden of proof on an inmate, RLUIPA throws the burden on prison officials, see id. § 2000cc-1(a). It is hard to imagine a greater reversal of fortunes for the religious rights of inmates than the one involved in the passage of RLUIPA.

240 F.Supp.2d at 575.

RLUIPA's enhanced protection for religious rights might not violate the First Amendment requirement of neutrality if Congress had enacted RLUIPA based upon evidence that religious rights are at greater risk of deprivation in the prison system than other fundamental rights. The exemption in Amos, for example, had the effect of maintaining congressional neutrality toward religion because "Title VII's prohibitions on hiring or firing on the basis of religion had a much greater negative impact on the purpose and mission of a religious organization in comparison to the effect of the prohibitions on a secular institution." Madison, 240 F.Supp.2d at 577 n. 9. In contrast, Congress enacted RLUIPA "[i]n the absence of any proof that religious rights are more at risk in prison than other fundamental rights, with the knowledge that strict scrutiny is not required to protect the religious belief of prisoners under the Free Exercise Clause," Madison, 240 F.Supp.2d at 576, and with the knowledge that prisoners already have a remedy for violations of their constitutional rights. "Such an action, while labeled a neutral `accommodation,' is not in fact neutral at all, and the Court is not allowed to defer to the mere characterization of RLUIPA as such." Id. at 576; see also Ghashiyah, 250 F.Supp.2d at 1027 ("The effect [of RLUIPA], therefore, is to provide greater protection to religiously motivated conduct than other conscientious conduct.").

Although the supporters of RLUIPA stated that "some institutions restrict religious liberty in egregious and unnecessary ways," see Statements of Senators Hatch and Kennedy, 146 Cong. Rec. S7774-01, S7775 (2000), RLUIPA supporters offered no evidence that inmates' other constitutional rights "are not similarly threatened by prison administrators," Madison, 240 F.Supp.2d at 575. And if prison officials in fact "restrict religious liberty [or other fundamental rights] in egregious and arbitrary ways," prisoners already have a remedy under Turner and O'Lone, which require prison policies to be "legitimate and neutral," Turner, 482 U.S. at 90, 107 S. Ct. 2254, and which held that strict scrutiny is not necessary to protect the religious rights of prisoners. Ghashiyah, 250 F.Supp.2d at 1031 (noting that O'Lone already provides prisoners a remedy for violations of their religious rights).

Madison provides an excellent illustration of the effect of RLUIPA upon the rights of prisoners:

Assume, for example, that a prison official confiscates white supremacist literature held by two different inmates. One inmate is a member of the Aryan Nation solely because of his fanatical belief that a secret Jewish conspiracy exists to control the world. The second inmate holds the white supremacist literature because he is a member of the Church of Jesus Christ Christian, Aryan Nation ("CJCC"). The non-religious inmate may challenge the confiscation as a violation of his rights to free expression and free association. A court would evaluate these claims under the deferential rational relationship test in Turner, placing a high burden of proof on the inmate and leaving the inmate with correspondingly dim prospects of success. However, the religious inmate, as a member of the CJCC, may assert a RLUIPA claim, arguing that the confiscation places a substantial burden on his religious exercise. The religious white supremacist now has a much better chance of success than the non-religious white supremacist, as prison officials bear the burden of proving that the prison policy satisfies a compelling interest and is the least restrictive means of satisfying the interest. The difference in the level of protection provided to each claim lies not in the relative merits of the claims, but lies instead in the basis of one claim in religious belief.

240 F.Supp.2d at 576 (internal citations omitted). As this example illustrates, the primary effect of RLUIPA is not simply to accommodate the exercise of religion by individual prisoners, but to advance religion generally by giving religious prisoners rights superior to those of nonreligious prisoners. "When Congress acts to lift the limitations on one right while ignoring all others, it abandons neutrality towards these rights, placing its power behind one system of belief. When the one system of belief protected is religious belief, Congress has violated the basic requirement of neutrality embodied in the Establishment Clause." Madison, 240 F.Supp.2d at 577 (internal citations omitted).

In addition to its message of endorsement, RLUIPA also has the effect of encouraging prisoners to become religious in order to enjoy greater rights. The Supreme Court has considered a statute's effect on nonreligious persons as part of the effect analysis. See Texas Monthly, Inc. v. Bullock, 489 U.S. 1, 9, 109 S. Ct. 890, 103 L. Ed. 2d 1 (1989) (plurality opinion) (holding that government may not compel nonadherents to support religious practices). One effect of RLUIPA is to induce prisoners to adopt or feign religious belief in order to receive the statute's benefits. As Ghashiyah explained:

[W]hen inmates see that the rules do not apply with the same force to the religious as to the agnostic or atheist ..., non-religious prisoners will know what they have to do so that they, too, can benefit from the softer rules: become religious. Considering the meager resources and opportunities available to them inside prison walls, the compulsion to become religious — created by government — will indeed be strong.

250 F.Supp.2d at 1029.

In evaluating a statute's effect, a court must ask "whether an objective observer, acquainted with the text, legislative history, and implementation of the enactment would view it as state endorsement of religion." Adland v. Russ, 307 F.3d 471, 484 (6th Cir.2002) (internal quotation marks omitted). RLUIPA's legislative history, as previously discussed, offers no evidence that religious rights are at any greater danger of deprivation in prison than are other fundamental rights. As to implementation, RLUIPA's inevitable effect is to give greater freedom to religious inmates, and to induce nonreligious inmates to adopt a religion. An objective observer viewing RLUIPA's text, legislative history, and effect would therefore conclude that the Act conveys a message of religious endorsement.

Plaintiffs, however, point to the following statement from Amos: "For a law to have forbidden `effects' under Lemon, it must be fair to say that the government itself has advanced religion through its own activities and influence." 483 U.S. at 337, 107 S. Ct. 2862 (emphasis in original). Amos held that the exemption from Title VII did not run afoul of the effect prong of Lemon because the government was not an active participant; the statute simply allowed religious organizations to pursue their religious objectives. Id. Plaintiffs contend that RLUIPA has an identical effect. RLUIPA, according to plaintiffs, "does not itself promote or subsidize a religious belief or message; it merely frees religious groups and individuals to practice as they otherwise would in the absence of certain significant state-imposed burdens."

The problem with plaintiffs' argument is that the exemption from Title VII that was at issue in Amos simply restored the level of freedom that religious institutions enjoyed before Congress enacted Title VII. By creating the exemption, Congress arguably acted to maintain neutrality toward religion. RLUIPA, on the other hand, does not lift any affirmative burden on the exercise of religion. Instead, by enacting RLUIPA, Congress itself has advanced religion by giving religious prisoners a preferred status in the prison community.

3. The entanglement between government and religion created by RLUIPA

The third prong of the Lemon test prohibits an excessive entanglement of government with religion. Lemon, 403 U.S. at 613, 91 S. Ct. 2105. Ghashiyah held that RLUIPA's nebulous definition of religious exercise, see 42 U.S.C. § 2000cc-5(7)(A), creates an excessive entanglement "because it forces the states to become involved with, knowledgeable about, and exceedingly sensitive to the varied religious practices of their inmates. It also forces the federal courts to become involved in prison administration, an area that the Supreme Court has admonished judges to avoid." 250 F.Supp.2d at 1031.

Although Ghashiyah's reasoning is plausible, we question whether RLUIPA requires any greater interaction between government officials and religion than exists under present law. Assume, for example, that a prisoner who is a member of a nontraditional "religion" claims a First Amendment right to a special diet that is required by the prisoner's "religious" beliefs. Before reaching the underlying constitutional claim, a court must first determine (1) whether the "religious" beliefs are sincerely held, and (2) whether the prisoner's beliefs "constitute a religion within the meaning of the [F]irst [A]mendment." Africa v. Pennsylvania, 662 F.2d 1025, 1029-31 (3d Cir.1981) (holding that a prisoner's belief system was not a religion for purposes of First Amendment analysis). Deciding whether a specific act or practice qualifies as "religious exercise" under RLUIPA arguably creates no greater entanglement than deciding whether a particular belief system constitutes a "religion" under the First Amendment. However, because we have concluded that RLUIPA has the impermissible effect of advancing religion, we have no need to further explore the question of whether RLUIPA violates Lemon's entanglement prong.

C. The unpersuasive nature of the cases upholding RLUIPA and RFRA

The cases that have upheld RLUIPA and RFRA against constitutional attack have essentially relied on the rationale of Amos. See, e.g., Charles v. Verhagen, 2003 WL 22455960, No. 02-3572, slip op. at 12-13 (7th Cir. Oct. 30, 2003); Mayweathers v. Newland, 314 F.3d 1062, 1068-69 (9th Cir.2002). We believe that such reliance is misplaced. As we have already discussed, the exemption in Amos was arguably necessary to avoid an Establishment Clause violation. Amos, 483 U.S. at 336, 107 S. Ct. 2862; see also id. at 344, 107 S. Ct. 2862 (Brennan, J., concurring). But RLUIPA extends protection to religious exercise in prison far beyond what is required by the Establishment Clause; it imposes strict scrutiny where the Establishment Clause requires only a rational-relationship review. See Turner, 482 U.S. 78, 107 S. Ct. 2254, 96 L. Ed. 2d 64; O'Lone, 482 U.S. 342, 107 S. Ct. 2400, 96 L. Ed. 2d 282.

The exemption in Amos, moreover, was a narrowly tailored solution to the potential Establishment Clause problem created by Title VII's application to religious institutions. RLUIPA, on the other hand, does not address a particular burden on religious exercise, but instead exempts religious prisoners from many generally applicable prison regulations. See Hamilton, 1 U. Pa. J. Const. L. at 13-14. We believe that Professor Hamilton's comment that "[c]omparing RFRA to Amos is like comparing apples to oranges" is equally applicable when comparing RLUIPA to Amos. See id. This same point was expressed somewhat differently in Madison:

The difference between Amos and RLUIPA is, like all Establishment Clause cases, a question of degree. However, the difference in degree between the two is substantial, and congressional neutrality is the line that divides them. When Congress has acted to impose an affirmative burden on religion, it is necessary for Congress to remove that burden in order to retain a position of neutrality towards religious belief. However, when Congress acts to provide religious inmates, and only religious inmates, with a level of constitutional protection that the Supreme Court has deemed unnecessary to protect religious rights, it has gone beyond protecting religion to affirmatively advancing it.

240 F.Supp.2d at 577 n. 9; see also Ghashiyah, 250 F.Supp.2d at 1028-29 (noting the fundamental distinction between the exemption in Amos and RLUIPA).

These authorities have convinced us that reliance on the rationale of Amos to sustain the constitutionality of RLUIPA is misplaced. We therefore conclude that the cases supporting RLUIPA are unpersuasive.

For all the reasons set forth above, we hold that 42 U.S.C. § 2000cc-1 violates the Establishment Clause. Because of this determination, we have no need to consider the alternative grounds raised by defendants in their constitutional challenge to RLUIPA. We therefore REVERSE the district court's denial of defendants' motions to dismiss and REMAND the case for further proceedings consistent with this opinion.