concurring in part and dissenting in part.
I join the majority opinion except for Sections II.A and II.B. Because I disagree with the majority’s conclusions regarding Fegans’s RLUIPA claims and would reverse and remand, I respectfully dissent.
RLUIPA requires strict scrutiny. 42 U.S.C. § 2000cc-l(a). While I agree with the majority that “context matters” in the application of RLUIPA’s strict-scrutiny standard and that we must accord “due deference to the experience and expertise of prison and jail administrators” in applying RLUIPA, ante at 902 (quotation omitted), I believe the majority accords Norris’s conclusory justifications a level of deference that is contrary to RLUIPA. Under RLUIPA, prison officials have the burden of establishing that their policy is the least restrictive means to achieve a compelling government interest. Id.; see also Hamilton v. Schriro, 74 F.3d 1545, 1552 (8th Cir.1996) (noting that under RFRA, “prison officials bear the burden of demonstrating that the regulation is the least restrictive means of achieving a compelling interest”). Norris has the burden of demonstrating that the grooming policy is the least restrictive means to achieve *909security as applied to Fegans. See 42 U.S.C. § 2000cc-l(a) (providing that prisons cannot “impose a substantial burden on the religious exercise of a person residing in ... an institution ... even if the burden results from a rule of general applicability” (emphasis added)); see also Gonzales v. O Centro Espirita Beneficente Uniao do Vegetal, 546 U.S. 418, 430-31, 126 S.Ct. 1211, 163 L.Ed.2d 1017 (2006) (determining RFRA requires individualized review); Roger v. Bryan, 523 F.3d 789, 796 (7th Cir.2008) (noting that RLUI-PA, unlike the Free Exercise Clause of the Constitution, requires individualized review); Spratt v. R.I. Dep’t of Corr., 482 F.3d 33, 39, 40 n. 9, 41 (1st Cir.2007) (rejecting the prison’s “all or nothing” argument and finding that the prison “must ... establish that prison security is furthered by barring [the individual] from engaging in” the disputed conduct); Washington v. Klem, 497 F.3d 272, 285 (3d Cir.2007) (noting that RLUIPA requires an assessment of whether the government action is the least restrictive means as applied to the individual). Norris must not only justify the policy as a whole, but he must also justify his failure to grant a religious-based exception to Fe-gans.
In upholding the grooming policy, the majority relies on Hamilton, a case where we upheld a prison’s hair-length regulation. Ante at 902-06. In Hamilton, however, we upheld the regulation based on a record that was “fully developed as to the prison officials’ basis for denying the inmate’s requests.... ” Pounders v. Kempker, 79 Fed.Appx. 941, 943 (8th Cir.2003) (unpublished) (citing Hamilton, 74 F.3d at 1547-48 and noting that the decision was based on a “fully developed” record detailing the prison officials’ decision). In this case, the record indicates Norris never even considered granting Fegans a religious-based exception, and the record is certainly not “fully developed as to [Norris’s] basis for denying [Fegans’s] request ]”. (T. Tr. 252). Also, in Hamilton, we explicitly distinguished a case where we struck down a hair-length regulation because “the only reason advanced in support of the regulation was the Warden’s opinion, unsupported by empirical proof,” that another alternative would be unworkable. Hamilton, 74 F.3d at 1555 n. 11 (distinguishing Teterud v. Burns, 522 F.2d 357, 361 (8th Cir.1975)). Furthermore, contrary to the facts of this case, the inmate in Hamilton was in a maximum security facility when he was denied the religious-based exception, 74 F.3d at 1547, and the Hamilton court was not faced with claims that evidence, such as a disparity in grooming policies among males and females or a medical exception, indicates that the grooming regulation could not possibly be the least restrictive alternative, see id. at 1554-55.
Our decision in Murphy v. Mo. Dep’t of Corr., 372 F.3d 979, 989 (8th Cir.2004), a case decided under RLUIPA, is instructive. We remanded Murphy, noting that the prison did not demonstrate that it “seriously considered any other alternatives” to a regulation alleged to infringe on prisoners’ free exercise of religion. 372 F.3d at 989. The prison denied a request by members of the Christian Separatist Church Society, a religion that believes Caucasians “are uniquely blessed by God,” to engage in group worship. Id. at 981-82. Prison officials testified that they denied the request, fearing the worship services would result in racial violence. We acknowledged in Murphy, and I acknowledge here, that prison officials’ security concerns are valid, id. at 988; nevertheless, under “RLUIPA’s higher standard of review, prison authorities must provide some basis for their concern that [safety problems] will result from any accommodation of [the] request.” Id. at 989.
*910In justifying policies under RLUIPA, “prison authorities must do more than offer conclusory statements and post hoc rationalizations for their conduct.” Hamilton, 74 F.3d at 1554 n. 10 (citing S.Rep. No. 103-111, at 10 (1993), as reprinted in 1993 U.S.C.C.A.N. 1892, 1900). Prison officials cannot “justify restrictions on religious exercise by simply citing to the need to maintain order and security in a prison. They ... must demonstrate that they actually considered and rejected the efficacy of less restrictive measures before adopting the challenged practice.” Alvarez v. Hill, 518 F.3d 1152, 1156 (9th Cir.2008) (citations and quotations omitted).
The majority does not, and can not, dispute that Norris must conduct an individualized review and that strict scrutiny applies to Norris’s decision not to grant a religious-based exception to Fegans. Instead, the majority opinion indicates that Norris did conduct an individualized review and that he determined that the grooming policy as applied to Fegans was the least restrictive means to achieve security. Ante at 905-08. However, the majority does not point to any evidence that prison officials considered whether to apply the policy to Fegans. Instead, the majority points to evidence that prison officials considered alternatives when formulating the policy as a whole. Ante at 903 n. 3. The majority opinion also does not cite to any evidence that Norris considered granting Fegans an exception. Norris testified he was “sure [prison officials] considered other alternatives” to the grooming policy that would achieve security; however, he could not recall any of these alternatives, nor remember how long they were considered, (T. Tr. 240), although he did testify that none of the alternatives involved a religious exception, (T. Tr. 235). Regardless of whether Norris’s failure to recall was because he forgot or because he never considered other alternatives, the record does not include any examples of less restrictive alternatives Norris considered when the policy was formulated or when declining to grant Fe-gans an exception. In fact, Norris explicitly testified that he would not consider granting a religious-based exception to anyone, regardless of the sincerity of their beliefs, (T. Tr. 252), and the prison warden testified he would not have considered granting one to Fegans, (T. Tr. 467). Considering all the evidence the district court credited, there is still no indication that prison officials seriously considered any alternatives to the grooming policy that would not substantially burden Fegans’s religious exercise. Thus, as a matter of law, I do not believe the policy, can be upheld under RLUIPA. See Murphy, 372 F.3d at 989 (remanding because the record did not indicate the prison “seriously considered any other alternatives, nor were any explored before the district court”).
The majority opinion concludes that “prison officials were aware” of his disciplinary history “[i]n applying this policy to Fegans.” Ante at 907. While an inmate’s disciplinary history may lead prison officials to conclude that security can only be achieved by applying a policy without exception to that inmate, the record does not reflect that prison officials considered Fe-gans’s disciplinary history in declining to grant him an exception. On the contrary, as indicated above, Norris testified that he would not grant anyone a religious-based exception. (T. Tr. 252). Furthermore, although Fegans had problems with discipline in the past, he was not in a maximum security facility when the grooming policy went into effect and Fegans was denied an exception. Ante at 900-01. Fegans’s placement at the time indicates prison officials did not believe he warranted the heightened restrictions of a maximum security facility. Even if prison officials *911were aware of Fegans’s disciplinary history, there is no evidence they considered it, or any other individual characteristic, in “applying] this policy to Fegans.”
In distinguishing the record in the instant case from the record in Warsoldier v. Woodford, 418 F.3d 989 (9th Cir.2005), the majority mischaracterizes Warsoldier and reverses the statutorily mandated burden of proof. Ante at 904-05. Warsoldier does not support the majority’s holding. In Warsoldier, the Ninth Circuit granted a preliminary injunction because the prisoner was likely to succeed on the merits of his claim that a gender-based grooming regulation, similar to the one in the instant case, violated RLUIPA. Id. at 1001. The court stated that the prison presented “only conclusory statements that the hair grooming policy is the least restrictive means to ensuring prison security” in its “attempt[ ] to meet its burden,” id. at 998, and that the data did not support the prison’s conclusions, id. at 1000-01. The court rejected the prison’s argument that the “court must completely defer to [the prison’s] judgment.” Id. at 1001. The court noted, as our court did in Murphy, 372 F.3d at 989, that the prison “cannot meet its burden to prove least restrictive means unless it demonstrates that it has actually considered and rejected the efficacy of less restrictive measures before adopting the challenged practice.” Warsoldier, 418 F.3d at 999. Even though the prison attempted to justify the gender-based differences by asserting that “women inmates are ‘much less likely’ to commit violent crimes than male inmates and, hence, that women inmates pose a lesser security concern,” the court found the policy violated RLUIPA. Id. at 1000.
However, instead of acknowledging that Warsoldier supports a conclusion that the prison policy in the instant case violates RLUIPA, the majority distinguishes War-soldier by putting the burden on Fegans to demonstrate that males do not pose a greater security risk than females. Ante at 905. I do not believe it is necessary for Fegans to provide “data to refute Norris’s expert testimony about the relative security risks in the male and female facilities,” ante at 905, because under RLUIPA, Fe-gans does not have the burden of proof. Norris does. 42 U.S.C. § 2000cc-1(a); Hamilton, 74 F.3d at 1552 (noting that under RFRA, “prison officials bear the burden of demonstrating that the regulation is the least restrictive means of achieving a compelling interest”). While the Ninth Circuit in Warsoldier did cite data to refute the prison officials’ concluso-ry assertions, this was data provided by the prison itself. The court gave no indication that the data was necessary or that it was the plaintiffs burden to rebut the prison officials’ assertions. Warsoldier, 418 F.3d at 1000. On the contrary, the court repeatedly emphasized that under RLUIPA, the prison, and not the plaintiff, had the burden to establish that the policy was the least restrictive means. Id. at 998, 1001; see also Hamilton, 74 F.3d at 1552. Regardless of whether the prison offers “a formal statistical presentation,” ante at 905 n. 2, conclusory assertions alone cannot, as a matter of law, satisfy Norris’s burden. See Hamilton, 74 F.3d at 1554 n. 10 (citing S.Rep. No. 103-111, at 10 (1993), as reprinted in 1993 U.S.C.C.A.N. 1892,1900).
Additionally, I do not believe Norris has adequately justified his failure to grant Fegans a religious-based exception in light of the fact that the policy requires of Fe-gans what it does not require of females and those with a certain medical condition. The majority accepts Norris’s conclusory statements to justify the grooming policy’s gender-based differences and medical exception: “ ‘Women are not generally as *912violent as men.... They are not as prone to give us problems with contraband as men.’ ” Ante at 905 (quoting (T. Tr. 237-38)). Norris also justifies the gender-based differences in the policy by stating that the differences were implemented “to let women be as much ... a woman as they can and still maintain good security.” (T. Tr. 237). Norris’s testimony is unsupported by any evidence, empirical or otherwise. His testimony indicates the policy was based on “generalizations or tendencies” about the differences between men and women, which are impermissible justifications even under intermediate scrutiny. United States v. Virginia, 518 U.S. 515, 541, 550-56, 116 S.Ct. 2264, 135 L.Ed.2d 735 (1996) (internal quotation omitted) (holding that a gender-based policy justified on assumptions about the differences between females and males does not survive intermediate scrutiny). While the prison’s gender-based policy is not subject to strict scrutiny, Norris’s failure to apply the policy equally between males and females and his conclusory and unsupported justification of the policy’s gender-based differences lead me to conclude the policy as applied to Fegans is not the least restrictive means to achieve safety and security.
Likewise, Norris’s willingness to grant an exception to those with a dermatological condition indicates the policy is not the least restrictive alternative. The majority opinion points out that “there may be good reason to distinguish” between Fegans and those with a dermatological condition. Ante at 907 n. 4. I do not dispute that there may be good reasons to provide exceptions to some and not to others. However, it is the prison’s burden to establish that the policy as applied to Fegans is the least restrictive means. Norris did not explain why this exception would not impair security but a religious-based exception would. He also did not explain why he granted this exception to “take good care of the inmates,” (T. Tr. 263), but he would not consider granting a religious-based exception. The prison has failed to establish this burden, and I do not believe it can be established by post-hoc rationalizations provided by the court.
While the majority points out that the existence of a medical exception does not “win the day for Fegans,” ante at 906, I believe that the medical exception, like the gender-based differences, indicates that exceptions can be made without sacrificing safety and security concerns. Norris justified his failure to accommodate Fegans by explaining that he would not allow any religious-based exceptions to the policy: “You have to treat everybody the same.... And if you don’t ... there wouldn’t be a grooming policy.” (T. Tr. 252). However, the prison’s grooming policy itself does not treat Fegans the same as females and those with a certain dermatological condition, thus undercutting Norris’s testimony that the policy must be uniformly applied in order to be effective. Just as prison officials considered whether security could be achieved while “let[ting] women be as much ... a woman as they can” and while “tak[ing] good care” of those with a dermatological condition, RLUIPA requires prison officials to consider whether security could be achieved while not substantially burdening Fegans’s exercise of his religion. The record indicates the prison did not consider this. It was thus improper for the district court to uphold the regulation as applied to Fe-gans. I would remand to give the prison the opportunity to conduct the individualized review RLUIPA requires.