Justice Clarence Thomas began Colorado’s half of the argument in St. Mary Catholic Parish v. Roy with a pointed observation. In the court’s last two cases about religious schools and public benefits, Espinoza v. Montana Department of Revenue and Carson v. Makin, the justices “did not use, even cite, Smith.” So, he asked Colorado’s deputy solicitor general, “Why would we do that in this case?”
That question goes to the heart of the case. After argument, it would be surprising if Colorado won. The real fight is over how the Catholic preschools win. And the answer will determine how much is left of Employment Division v. Smith, perhaps the court’s most contested free exercise precedent.
Some background
In 2020, Colorado voters approved funding for universal preschool. The program that followed pays for 15 hours a week of preschool for every four-year-old in the state, worth about $6,300 per child. Families can spend that money at roughly 2,000 participating preschools: public, private, and religious. Every provider must accept an “equal opportunity” mandate. Specifically, children must have an equal opportunity to enroll “regardless of race, ethnicity, religious affiliation, sexual orientation, gender identity, lack of housing, income level, or disability.”
The mandate is not absolute. Colorado lets preschools use enrollment “preferences” for, among other things, children of employees, siblings, members of a specific community, and families willing to volunteer. More than 1,000 preschools use at least one. But Catholic parish preschools in the Archdiocese of Denver ask enrolling families to support the Church’s teachings on sex and marriage, and they treat children according to their biological sex. Families must affirm those teachings in good faith. Colorado says those practices violate the mandate, so the parishes cannot participate. Families who want a Catholic preschool lose the benefit. Petitioners Daniel and Lisa Sheley have lost more than $12,000 so far. The lower courts largely sided with Colorado.
The legal backdrop
The free exercise clause bars the government from “prohibiting the free exercise” of religion. For decades, the court read that guarantee to require strict scrutiny of laws that substantially burden religious practice. In 1963’s Sherbert v. Verner, South Carolina denied unemployment benefits to a Seventh-day Adventist who would not work on her Sabbath. The court held that the state could not force her to choose between her faith and a public benefit without a compelling reason.
Smith changed course in 1990. Oregon had denied unemployment benefits to two men fired for using peyote in a Native American Church ceremony. Writing for the court, Justice Antonin Scalia held that a neutral law of general applicability (that is, one that neither targets religious conduct nor exempts comparable secular conduct) needs only a rational basis, however heavily it burdens religious exercise. Under that rule, the government almost always wins.
But Smith did not overrule Sherbert. It distinguished it, and the court has spent the decades since narrowing Smith’s reach. Petitioners count eight categories of free exercise cases in which the court has declined to apply it: animus toward a religious group or person (Masterpiece Cakeshop); church autonomy (Hosanna-Tabor); discrimination among religions (Larson v. Valente); forced participation in ceremonies (Barnette); hybrid rights, which pair free exercise with another constitutional right (Cantwell); parental rights over children’s religious upbringing (Mahmoud v. Taylor); penalties on religious belief as such (McDaniel v. Paty); and the receipt of public benefits (from Sherbert through Carson). Even within Smith’s domain, a law gets strict scrutiny if it is not truly “neutral” or “generally applicable,” for example because it allows individualized exemptions (Fulton) or treats comparable secular conduct more favorably (Tandon).
The public benefits category is where Carson comes in. In three cases over five years, the court held that states may not exclude religious schools from otherwise available funding. 2017’s Trinity Lutheran involved a church preschool barred from a playground-resurfacing grant. 2020’s Espinoza involved a scholarship program closed to religious schools. And in Carson, decided in 2022, Maine paid tuition at private schools but excluded schools that taught religion. The court said all of these limits violated the free exercise clause. “Regardless of how the benefit and restriction are described,” it explained, “the program operates to identify and exclude otherwise eligible schools on the basis of their religious exercise.” Its decisions, the court said, “turned on the substance of free exercise protections, not on the presence or absence of magic words.”
Two paths to the same result
The Catholic preschools offer the court two routes to strict scrutiny in this case. They track the two questions the court agreed to decide.
The first route stays inside Smith. Colorado’s mandate, petitioners argue, is not generally applicable. The state lets more than 1,000 preschools limit enrollment for secular reasons but refuses to accommodate the parishes’ religious ones. The U.S. Court of Appeals for the 10th Circuit rejected that argument, holding that a challenger must show unfettered discretion or categorical exemptions for identical secular conduct.
The second route goes around Smith. Under Carson, the preschools argue, strict scrutiny applies whenever the government conditions a generally available benefit on giving up religious exercise, whether or not the condition is neutral. Colorado, for its part, reads Carson far more narrowly, as a rule against excluding schools because they are religious. Colorado does not exclude religious preschools as such, and many participate. On that view, Carson does not apply and Smith does.
The choice between those routes is where the real stakes lie. A win on general applicability would be narrow and fragile. It would turn on the details of Colorado’s exemptions. Colorado, or any other state, could respond by trimming its secular exemptions and excluding the same schools again. The next challenger would have to start over, hunting for a new secular comparator. Principal Deputy Solicitor General Sarah Harris called that regime “an endless sort of whack-a-mole game.”
A win under Carson, in contrast, would be far more durable. It would hold that conditioning a public benefit on abandoning religious exercise triggers strict scrutiny, however the condition is drafted. That rule would stop the kind of maneuvering Maine engaged in after Carson itself, when it rewrote its program to exclude the same religious schools without using religious labels. And it would take an entire domain, public benefits, out of Smith’s reach.
That is why Thomas’ question, which this article began with, mattered. Going in, some commentators wondered whether the court would use this case to bury Smith. CNN’s Joan Biskupic framed it that way. The argument suggested something more measured, though still significant. If the court does not overrule Smith here, it seems poised to quarantine it as far as it can.
Here’s one version of what that quarantine could look like. The court can read Carson’s public benefits rule for what it says. It can apply strict scrutiny, properly understood, in that domain. It can ground that choice in the clause’s original meaning. And it can say plainly what the United States told the court: Smith is the anomaly.
Carson means what it says
Justice Neil Gorsuch asked what Colorado’s narrow reading would leave of a decade of precedent. On Colorado’s view, he wondered, had the court been “just wasting our time for the last 10 years” on Carson,Trinity Lutheran, and Espinoza? If so, any state could escape those cases by redrafting.
Indeed, states are doing exactly that. Petitioners’ counsel, Nick Reaves, told Gorsuch that “before the ink was even dry in Carson, Maine changed its laws” so that they “don’t explicitly call out religion, but they still exclude the same families and the same schools.” Colorado’s own record shows the same pattern. The state amended its discretionary enrollment exemption process during trial. More than once, it redefined an exemption that originally allowed faith-based preschools to limit enrollment to members of their congregation. Then, while the case was on appeal, it repealed that congregation exemption entirely and argued that the repeal freed it from the district court’s injunction.
Reading Carson for what it says would not create a boundless exception. Reaves located the boundary in Smith’s own concern. Smith “was concerned about every man becoming a law unto himself,” he said, and about “broad exemptions to criminal prohibitions. That interest is not as strong when it comes to public benefits when the underlying conduct is lawful.” When Justice Amy Coney Barrett asked whether a law making the mandate a condition of every preschool’s license to operate would fall under Carson, Reaves conceded, “I don’t think that would be a Carson case.”
Harris marked a second boundary. Justice Sonia Sotomayor asked about Bowen v. Roy, a 1986 decision rejecting a Native American father’s religious objection to the government’s own use of his daughter’s Social Security number. Harris pointed to the part of Bowen that commanded a majority: “You can’t force a change in the way in which the government itself manages internal operations.” Government conduct that neither pressures nor prohibits religious exercise triggers no scrutiny, as Michalyn Steele and I have explained elsewhere about Bowen. In sum, Carson’s rule has limits on both sides. It reaches conditions on public benefits for lawful conduct, not licensing regimes or the government’s management of its own affairs.
The sky will not fall
The natural worry is what comes next. If strict scrutiny governs benefits conditions, will every program become a negotiation with every objector? Justice Elena Kagan pressed this concern. Everything, she said, would turn on whether a law passes “strict scrutiny, which is a kind of balancing test, which is like how important is the state interest? How important is the religious belief? How do we compare the one to the other?”
But that is a caricature of strict scrutiny, at least as the court applies it in this context. Gorsuch offered the more accurate account, and Harris confirmed it. “[W]e assess whether the government has a compelling interest and whether it’s narrowly tailored to that interest without balancing against the interest on the other side,” Gorsuch said. Once a claimant shows a substantial burden, “[w]e don’t go balancing the two.” Gorsuch is exactly right about that.
I do not deny that older strict scrutiny opinions sometimes used the vocabulary of balancing. But the doctrine has since been disciplined, largely through the courts’ experience with the Religious Freedom Restoration Act and the Religious Land Use and Institutionalized Persons Act, passed in Smith’s wake. Courts applying RFRA and RLUIPA do not ask how central a belief is to a faith. They ask a threshold question about whether the government has objectively burdened religious exercise. At the scrutiny stage, they do not weigh the value of religion against the value of a policy. As I have argued elsewhere, strict scrutiny works as an exclusionary norm paired with a heavy evidentiary burden. It excludes all but a narrow set of permissible reasons for burdening religious exercise. Then it requires the government to prove that its action is necessary to serve one of them. That’s precisely what Gorsuch described.
The court has also adopted this disciplined approach in other contexts. In United States v. Hemani last term, it held that the government cannot disarm a person merely by naming a permissible reason. It must prove that the reason is one it actually acts on, and tie that reason to the person. As I wrote at the time, history fixed the permissible ends, and the court then demanded proof that the means served them. That is how constitutional rights work more generally: as constraints on the reasons government may act on. Strict scrutiny does not ask how much religion matters. It asks whether the state’s reason is one the Constitution permits, and whether the state can prove that reason applies to this claimant.
Harris invoked Gonzales v. O Centro, the court’s leading RFRA strict scrutiny precedent, for this very point. “I think O Centro rightly recognized strict scrutiny,” she said. “[E]ven when people say it’s tough, it is a very manageable standard for courts. It’s a familiar standard for courts.” O Centro requires the government to show that its interest is served by applying the law “to the person,” meaning “the particular claimant whose sincere exercise of religion is being substantially burdened.” Courts must look “beyond broadly formulated interests” to the harm of granting the specific exemption requested.
The track record bears this out. Gorsuch noted that the federal government “already has to meet strict scrutiny” in providing its many public benefits. Harris pointed to the federal experience under RFRA and to “the 40 states that have RFRA equivalents.” Under the pre-Smith regime, she added, cases “turned out both ways.” Public schools already operate under this regime. The District of Columbia’s schools are subject to federal RFRA. As Gorsuch put it, Senators Ted Kennedy and Orrin Hatch “would be quite surprised to think that RFRA created laws unto themselves in every American.”
Smith, by contrast, is the unstable regime. The court’s cases have left what Harris called “a mess,” in which “it’s hard to describe Smith as a default, and it’s even harder to describe Smith as a workable default in this particular context.” Strict scrutiny is the familiar standard. Smith, in her words, is “a much less stable jurisprudence.”
Closer to the original meaning
Restoring strict scrutiny for benefits conditions would also move free exercise doctrine closer to the clause’s original meaning. Two justices asked whether strict scrutiny is faithful to that meaning, and both counsel answered yes.
Barrett asked Reaves whether strict scrutiny is “what you think is required by the original meaning.” Reaves answered that Sherbert v. Verner is “much closer to the original meaning than Smith.” He pointed to the “peace and safety provisos” in founding-era state constitutions, which protected religious exercise unless it threatened public peace or safety. Barrett then put the point precisely: “So you see it as an implementing doctrine that’s close to how it would have been implemented at the time of the founding?” “I think that’s right, yes,” Reaves said. Gorsuch asked Harris a parallel question. Is strict scrutiny, as the court announced it in Sherbert and Thomas v. Review Board and applied it in the recent benefits cases, “a reasonable construction of … the original meaning?” Harris said yes.
That framing is right. Strict scrutiny is not itself the original meaning. It is a doctrine for implementing that meaning, a form of what originalists call constitutional construction. Mark Storslee and I made this argument in our amicus brief in the case, which petitioners’ counsel also cited at argument for the history of Catholic exclusion from the common schools. The brief draws on my work elsewhere, including Constructing Constitutional Rights and Replacing Smith.
The core of the argument is this. The founding generation did not let religious exercise yield to any reason government happened to articulate. Early state constitutions took some justifications for burdening religious exercise off the table. Others, most commonly the protection of public peace and safety, were permissible. And founding-era courts made the government show that it was actually advancing one of those permissible interests before religious exercise had to give way. For example, in People v. Philips, decided in 1813, a New York court refused to force a Catholic priest to testify about a confession. Compelling him, the court explained, would place him “between Scylla and Charybdis,” and the priest’s silence posed no threat to the peace or safety the state constitution protected. That is what strict scrutiny requires: a permissible reason, proven on the facts.
The evidence from Reconstruction points the same way. The 14th Amendment, which applies the free exercise clause to the states, was framed by a generation shaped by religious resistance to the Fugitive Slave Act. Abolitionist believers insisted on a right to give a fugitive shelter and a “crust of bread,” whatever neutral and generally applicable law said. As Kurt Lash and I have explained, that understanding informed the amendment’s privileges or immunities clause. Judge Ryan Nelson of the 9th Circuit relied on that history in a recent concurrence concluding that Smith “breaks from the history and tradition surrounding free exercise and the Fourteenth Amendment.” At argument, Harris agreed with Justice Brett Kavanaugh that, as a matter of original meaning and contrary to Smith, the free exercise clause is “more than simply a protection against explicit discrimination.”
The anomaly is Smith
One serious question about the Carson approach came from Kavanaugh. A rule that applies strict scrutiny to benefits conditions, but not to outright bans, would impose “greater First Amendment scrutiny on a benefits program than … an outright ban on the religious conduct.” That, he said, “seems inverted.” On petitioners’ theory, a state that merely withholds funding would face strict scrutiny, while a state that banned the same conduct outright, through a neutral law, would get only Smith’s rational basis review. He acknowledged one answer, that the court “should just overrule Smith entirely,” but asked Reaves to assume that was off the table.
The incoherence is real. And if the court doesn’t overrule Smith in this case, hopefully it will take that step someday soon. But the incoherence belongs to Smith itself. Harris put it directly: “Smith is the anomaly in the context of public benefits or in any other context, where other rights do not have this so long as a law is generally applicable rational basis review.”
Smith itself claimed the opposite, calling religious exemptions “a constitutional anomaly.” As Mark Rienzi and I have argued, the reverse is true. Exemptions are a species of as-applied challenge, the default remedy elsewhere in constitutional law. The anomaly is Smith’s rule, not the exemptions it disfavors.
The court has retired other precedents it no longer believes in. It distinguishes them, whittles them away, and finally overrules them. Teague v. Lane’s “watershed” exception for new rules of criminal procedure went unused for 32 years before Edwards v. Vannoy declared it “moribund” and eliminated it. Gorsuch joined that opinion. Abood v. Detroit Board of Education was cabined inHarris v. Quinn before Janus v. AFSCME overruled it. Lemon v. Kurtzman was sidestepped for years before Kennedy v. Bremerton confirmed that the court had “long ago abandoned” it.
After decades of whittling, quarantine is the penultimate stage of that life cycle. The court can further limit Smith’s damage now. Eventually the court will likely complete the life cycle and retire Smith altogether.
Why are we here?
The most striking moments in the argument were strict scrutiny questions in all but name. Justice Samuel Alito asked how many times the feared harm had actually happened, and observed that there was “little evidence of a practical problem.” Barrett asked whether there was any record of LGBTQ parents being turned away from these preschools, and whether any entity other than religious ones had sought an exemption from the mandate. “No, Your Honor,” Colorado’s counsel answered. Kavanaugh asked the state, “Why not seek the win/win?” With roughly 2,000 preschools in the program and 34 Catholic ones, he saw “plenty of opportunities to respect the rights of the religious believers and respect the ability of LGBT parents to find” a school that fits. “Why has this been provoked?” he asked.
Those are the questions narrow tailoring forces a government to answer. Strict scrutiny asks whether the state can prove that its reason applies here, not merely name it. That demand makes room for the win-win Kavanaugh described. Religious preschools can serve the families who want them, and every family can find a school that fits. Smith invites the opposite: winner-take-all conflicts provoked by a rule that never asks whether the conflict was necessary.
The best case for quarantining Smith is not that religious claimants deserve to win. It is that strict scrutiny asks the questions this argument kept asking: Why are we here, and couldn’t we have avoided this conflict? A rule that makes the government answer those questions is easy to administer. It is also faithful to the clause’s original meaning.
Disclosure: The author was amicus curiae in support of the petitioners in St. Mary Catholic Parish v. Roy.
