Prayer Studies: What Controlled Trials Have Found

The question of whether prayer produces measurable effects on physical outcomes is one of the few theological claims that can be tested experimentally. Scientists have run those tests. The results deserve a careful look.

What researchers were trying to measure

Intercessory prayer — prayer offered by one person on behalf of another — became the focus of formal clinical trials beginning in the 1980s. The appeal from a methodological standpoint is obvious: if prayer influences health outcomes, a randomized controlled trial should be able to detect it. You assign patients to a prayed-for group and a control group, keep the patients blind to their assignment, standardize the prayer intervention, and measure clinical endpoints. This is the same basic architecture used to test drugs. The approach does require one significant assumption — that the effect, if it exists, is consistent enough to show up in aggregate data rather than being entirely at divine discretion — but proponents of prayer generally claim it is.

The most ambitious study in this area is the Study of the Therapeutic Effects of Intercessory Prayer (STEP), published in 2006 in the American Heart Journal. It enrolled 1,802 patients recovering from coronary artery bypass surgery across six hospitals. Three Christian groups offered intercessory prayer for patients by first name. The result: no significant difference in complication rates between prayed-for and not-prayed-for patients who did not know their assignment status. One group that was told with certainty that strangers were praying for them actually showed a slightly higher complication rate — a finding the authors attributed plausibly to performance anxiety rather than any causal effect of prayer.

The pattern across earlier and later trials

STEP was not an outlier. Earlier work by Randolph Byrd (1988) and William Harris (1999) had claimed to find positive effects of intercessory prayer in cardiac patients, and those studies drew significant attention. Both have since been criticized on methodological grounds: weak blinding procedures, multiple outcome measures without correction for multiple comparisons, and inconsistencies between the primary analyses and the reported results. When critics applied more rigorous standards, the apparent effects disappeared or shrank to statistical noise.

A 2007 Cochrane systematic review — the gold standard format for synthesizing clinical evidence — examined all available randomized trials of intercessory prayer. It identified ten trials meeting inclusion criteria. The reviewers concluded there was no clear evidence that prayer is an effective treatment for any health condition, and they noted that the methodological quality of available studies was generally poor. A 2009 update did not change that conclusion. More recent meta-analyses covering a broader set of distant healing practices, of which intercessory prayer is a subset, have consistently found small or null effects that track with study quality: the better the methodology, the smaller the effect.

How believers interpret these results

The most common theological response is that God cannot be tested — that treating prayer as an independent variable in a clinical trial misunderstands what prayer is. This is a serious objection worth engaging honestly. If prayer is fundamentally relational, an expression of alignment with divine will rather than a mechanism for producing outcomes, then a randomized trial may simply be the wrong instrument. A study cannot measure whether a relationship is authentic.

The problem is that this response is available after the fact in a way that undermines any empirical investigation. Before STEP was published, many practitioners and theologians were willing to say that positive results would confirm the reality of God's intervention. The prior willingness to count confirming evidence as evidence, combined with the post-hoc move to declare negative evidence inadmissible, is a recognizable pattern in motivated reasoning. It is not unique to religious believers — motivated reasoning appears across many domains of human belief — but it should be named clearly when it appears.

A more intellectually consistent position, held by some theologians, is to grant that intercessory prayer should not be expected to produce statistically detectable health effects and that the clinical trials were misconceived from the start. That position is at least coherent. It does, however, require walking back claims that many religious communities make in practice — that prayer heals, that God answers specific requests, that outcomes differ for those who pray.

What remains genuinely open

Prayer may have real psychological benefits for the person praying: reduced anxiety, increased sense of agency, social cohesion within communities of shared practice. Some of these effects have support in the literature and are worth taking seriously on their own terms. The question of whether those benefits require any metaphysical claim to be true is separate from whether the metaphysical claim is itself supported by evidence.

The controlled trials on intercessory prayer do not prove that no God exists. They test a narrower claim: that prayer of a specific type reliably changes physical outcomes in third parties. On that specific, testable claim, decades of reasonably well-conducted research have returned a consistent answer.