Prayer and Empirical Evidence

Introduction

The question of whether prayer produces measurable effects on the physical world has been a subject of empirical inquiry since at least the nineteenth century, when the polymath Francis Galton conducted what is often considered the first statistical analysis of prayer’s efficacy.⁶ Galton observed that despite the widespread practice of praying for the health of British monarchs, their longevity did not exceed that of other affluent demographics—a finding he regarded as evidence against prayer’s therapeutic power.⁶

The twentieth and twenty-first centuries saw the question taken up with the tools of modern clinical research: randomized controlled trials, double-blind protocols, and systematic reviews. The resulting body of evidence is substantial and, from the standpoint of empirical science, remarkably consistent. No major peer-reviewed study has demonstrated a statistically robust, replicable benefit of intercessory prayer—prayer offered by third parties on behalf of patients who may or may not be aware of it.⁵¹⁰

These findings raise profound questions at the intersection of philosophy of science, theology, and epistemology. If prayer is a genuine causal mechanism connecting human petition to divine intervention, why does it fail under controlled observation? And if it is not, what does that imply about the nature of prayer in religious tradition? This article surveys the major clinical studies, the theological responses they have provoked, and the philosophical issues that frame the debate.

Major Clinical Studies

The Byrd Study (1988)

The first major randomized controlled trial of intercessory prayer was conducted by cardiologist Randolph Byrd at San Francisco General Hospital. Byrd randomized 393 coronary care unit patients into a prayer group and a control group. Intercessors—“born again” Christians recruited from Protestant and Catholic congregations—were given patients’ first names, diagnoses, and general conditions, and asked to pray daily for rapid recovery and prevention of complications.¹

Byrd reported that the prayer group had statistically significantly fewer instances of congestive heart failure, fewer cases requiring diuretics, fewer episodes of pneumonia, fewer instances of intubation, and fewer cases of antibiotics usage compared to the control group.¹ The study attracted widespread media attention and was hailed by some religious commentators as scientific validation of prayer.

However, the study drew significant methodological criticism. The outcome variables were numerous (26 separate categories), increasing the probability of finding statistically significant results by chance alone. The study lacked a clear primary endpoint defined in advance, a standard requirement for clinical trials to prevent post hoc data mining. Critics also noted that there was no way to ensure that control-group patients were not also receiving prayer from their own friends and family, a confound inherent in all intercessory prayer studies.⁷¹⁸

Harris et al. (1999)

William Harris and colleagues at the Mid America Heart Institute attempted to replicate and extend Byrd’s findings. In a study of 990 coronary care unit patients, intercessors prayed daily for 28 days for patients assigned to the prayer group. Neither patients nor staff knew which group patients were assigned to. Harris used a weighted severity score (the Mid America Heart Institute Cardiac Care Unit scoring system) as the primary outcome measure.²

The prayer group showed a modest but statistically significant improvement on this weighted score (a 10% lower score). However, there was no significant difference in length of hospital stay, and no difference on a separate scoring system (the Byrd score) applied to the same data.² Critics argued that the weighted scoring system had been developed specifically for this study and that the results were not robust across different analytical approaches. The selective reporting of a favorable outcome measure, while a less favorable one showed no difference, raised concerns about outcome-switching bias.⁷¹⁰

The MANTRA Study (2005)

The MANTRA II (Monitoring and Actualisation of Noetic Trainings) study, led by Mitchell Krucoff at Duke University, took a broader approach. It enrolled 748 patients undergoing percutaneous coronary intervention (angioplasty) at nine medical centers and tested several “noetic therapies”—intercessory prayer, music therapy, imagery, and touch therapy—individually and in combination.³

The study found no significant effect for any noetic therapy, either individually or combined, on the primary composite endpoint of major adverse cardiovascular events at six months. Prayer groups showed no measurable benefit compared to standard care.³ The MANTRA II study was notable for its multi-site design and relatively large sample size, making its null result more difficult to dismiss as a statistical fluke.

The STEP Study (2006)

The Study of the Therapeutic Effects of Intercessory Prayer (STEP), led by Herbert Benson at Harvard Medical School and funded by the John Templeton Foundation at a cost of $2.4 million, remains the largest and most methodologically rigorous trial of intercessory prayer ever conducted.⁴¹⁷ The study enrolled 1,802 patients undergoing coronary artery bypass graft (CABG) surgery at six medical centers and randomized them into three groups:

Group 1: Patients who were told they might or might not receive prayer, and did receive prayer (604 patients).
Group 2: Patients who were told they might or might not receive prayer, and did not receive prayer (597 patients).
Group 3: Patients who were told they would definitely receive prayer, and did receive prayer (601 patients).⁴

Intercessory prayer was provided by members of three Christian groups: the Community of Teresian Carmelites in Worcester, Massachusetts; the Society of the Little Sisters of Jesus in Maryland; and Silent Unity, the prayer ministry of the Unity Church in Missouri. Each group was given patients’ first names and last initials and asked to include the phrase “for a successful surgery with a quick, healthy recovery and no complications” in their prayers over a 14-day period beginning the night before surgery.⁴

The primary outcome was any complication within 30 days of surgery. The results were striking and unambiguous:

Groups 1 and 2—those uncertain about whether they were receiving prayer—showed no significant difference in complication rates (52% in the prayer group vs. 51% in the no-prayer group). Prayer had no detectable therapeutic effect.⁴

Group 3—patients who knew with certainty they were being prayed for—had a significantly higher complication rate of 59%, compared to 52% in Group 1 (who also received prayer but did not know it for certain). This unexpected finding suggested a nocebo effect: the knowledge of being prayed for may have increased anxiety, perhaps because patients interpreted it as an indication that their condition was especially serious.⁴¹⁷

The STEP study was pre-registered, multi-site, adequately powered, and double-blinded for the comparison between Groups 1 and 2. Its null result carried particular weight because the Templeton Foundation, which funded the research, had a stated interest in finding evidence for the intersection of science and spirituality.¹⁷

The Cochrane Systematic Review

In 2009, the Cochrane Collaboration—the gold standard for evidence-based medicine reviews—published a systematic review of all available randomized controlled trials on intercessory prayer for the alleviation of ill health. The review, led by Leanne Roberts, identified ten studies involving 7,646 patients across various clinical contexts.⁵

The review’s conclusion was unequivocal: “The findings are equivocal, and although some of the results of individual studies suggest a positive effect of intercessory prayer, the majority do not, and the evidence does not support a recommendation either in favour or against the use of intercessory prayer.”⁵ The reviewers noted significant methodological heterogeneity across studies, including variations in prayer protocols, outcome measures, patient populations, and follow-up periods. They concluded that no firm conclusions could be drawn about the efficacy of intercessory prayer and raised the question of whether further trials were warranted, given both the consistent null results and the inherent methodological challenges.⁵

Meta-Analyses and Aggregate Findings

A comprehensive review by Masters and Spielmans in the Journal of Behavioral Medicine (2007) analyzed the accumulated research on prayer and health outcomes. They concluded that methodologically rigorous studies consistently failed to find significant effects of intercessory prayer, and that the occasional positive results in earlier studies were attributable to methodological weaknesses, including multiple comparisons, post hoc outcome selection, and inadequate blinding.¹⁰

The pattern across the literature is notable: as study quality and sample size increase, effect sizes diminish toward zero or become slightly negative. This is a well-known signature of a null effect contaminated by bias in smaller, less rigorous studies—a pattern observed across many areas of research where initial promising findings fail to replicate under more stringent conditions.¹⁰

Methodological Challenges

All studies of intercessory prayer face inherent methodological difficulties that complicate interpretation, regardless of their findings.

The most fundamental challenge is the contamination problem: there is no way to ensure that control-group patients are not also being prayed for by friends, family, church communities, or strangers. If the hypothesis is that prayer works, then a “no prayer” control group almost certainly does not exist in a society where religious practice is widespread. This confound could theoretically mask a real effect, though it could equally be taken as evidence that ambient prayer (which is presumably constant and universal for hospitalized patients in religious societies) produces no measurable benefit.⁷¹⁸

There is also the dosage problem: studies have no principled way to determine how much prayer constitutes a “therapeutic dose.” Unlike pharmacological trials where milligrams can be precisely measured, prayer varies in duration, intensity, sincerity, theological tradition, and the spiritual state of the intercessor. Proponents can always argue that the prayer offered in a given study was insufficient, of the wrong type, or offered by the wrong people.¹⁸

The specificity problem compounds the dosage issue: researchers must make arbitrary decisions about which religious tradition’s prayers to test, what instructions to give intercessors, and what outcomes to measure. If Christian prayer fails, a proponent might argue that Buddhist or Jewish prayer would have succeeded, or that the specific wording was wrong, or that God chose not to act because the context was a clinical trial.⁷

Richard Sloan and Emilia Bagiella have argued that these methodological difficulties are not merely practical obstacles but point to a deeper conceptual problem: the hypothesis that prayer heals may not be a scientific hypothesis at all, because it cannot be formulated in a way that specifies the conditions under which it would be falsified.⁷ This connects the empirical question to broader issues in the philosophy of science regarding demarcation criteria and the unfalsifiability of theistic claims.

Theological Responses

The negative findings of prayer research have prompted a range of theological responses, many of which argue that the studies are misguided not because of poor methodology but because of flawed theology.

God Is Not a Vending Machine

The most common theological objection holds that prayer is not a transactional mechanism for obtaining desired outcomes. To treat prayer as though it should produce measurable results on demand is to misunderstand its nature fundamentally. Prayer, on this view, is a relational act—a conversation with God, an expression of trust, an alignment of the human will with the divine will—not a technique for manipulating physical reality.¹⁶

This objection has deep roots in Christian theology. Jesus’s prayer in Gethsemane—“not my will, but yours be done” (Luke 22:42)—is frequently cited as the paradigmatic model of prayer in the Bible: genuine prayer submits to God’s purposes rather than attempting to override them. If God’s answer to prayer is sometimes “no,” then a clinical trial measuring only “yes” outcomes would miss the point entirely.

God Cannot Be Tested

A related objection draws on the biblical injunction against testing God. Deuteronomy 6:16 states: “Do not put the Lord your God to the test,” a passage Jesus himself quotes when refusing Satan’s challenge to throw himself from the temple pinnacle (Matthew 4:7). On this view, a randomized controlled trial of prayer is precisely the kind of demand for a sign that God has explicitly refused to honor. God, being a free agent and not a natural force, might deliberately withhold action when subjected to experimental testing, not because prayer is ineffective but because coerced demonstration would undermine the very faith that prayer presupposes.¹⁸

Critics have noted that this argument, while internally consistent, renders the prayer hypothesis unfalsifiable. If positive results would confirm prayer’s power but negative results are explained by God’s refusal to be tested, then no possible outcome could disconfirm the claim. Karl Popper identified unfalsifiability as the hallmark of a non-scientific hypothesis—not necessarily a false claim, but one that lies outside the reach of empirical investigation.⁸

Prayer Changes the Pray-er

A subtler theological response concedes that intercessory prayer may not produce measurable physical effects on distant patients, but argues that this was never its primary purpose. Prayer’s real function, on this account, is transformative for the person who prays: it cultivates compassion, reduces anxiety, strengthens community bonds, and fosters a sense of meaning and connection to the transcendent.¹⁶

William James, in The Varieties of Religious Experience, documented extensive evidence that prayer and other spiritual practices produce profound subjective changes in practitioners—shifts in attitude, emotional regulation, sense of purpose, and even physiological markers of calm.¹⁶ These effects are not disputed by the clinical literature; what the studies tested was the much stronger claim that prayer produces effects on people other than the one praying, at a distance, without any known physical mechanism.

This retreat from intercessory efficacy to personal transformation is theologically coherent but represents a significant narrowing of traditional prayer claims. If prayer changes only the pray-er, then petitionary prayer—asking God to intervene in specific situations—is at best a psychologically useful exercise rather than a genuine request directed at an agent who might act on it.

The Free Will Defense

Some theologians invoke a version of Alvin Plantinga’s free will defense to explain why prayer does not produce reliable empirical results.¹¹ On this view, God values human freedom so highly that systematic divine intervention in response to prayer would effectively coerce belief, undermining the conditions necessary for genuine faith. A world in which prayer reliably healed the sick would be a world in which disbelief was irrational—and therefore a world in which faith, in the theological sense of trust in the unseen, was impossible.

This argument parallels the “divine hiddenness” response to the problem of evil: God remains hidden (or at least ambiguous) precisely to preserve the epistemic space in which free creatures can choose to believe or not believe. The argument from miracles faces similar tensions, as any putative miracle that could serve as evidence for God’s existence would, on this logic, need to remain ambiguous enough to be denied by skeptics.

Philosophical Issues

Falsifiability and Demarcation

The prayer hypothesis illustrates a classic problem in the philosophy of science. For a claim to be scientifically testable, it must be possible to specify in advance what observations would count against it.⁸ Prayer advocates who accept positive studies as confirmation but dismiss negative studies as irrelevant (“God cannot be tested,” “the wrong kind of prayer was used,” “control patients were also being prayed for”) have effectively immunized their hypothesis against disconfirmation.

This does not mean the claim “prayer works” is false. It means it is not an empirical claim at all, in the Popperian sense. It belongs to the same category as other unfalsifiable theistic claims—assertions that may be meaningful within a theological framework but that cannot be adjudicated by the methods of natural science.⁸¹⁴

Confirmation Bias and Selective Memory

Outside the clinical setting, individual belief in prayer’s efficacy is powerfully sustained by confirmation bias—the well-documented human tendency to notice and remember evidence that supports existing beliefs while ignoring or forgetting evidence that contradicts them.¹⁴

When a believer prays for a sick relative who subsequently recovers, the recovery is attributed to prayer. When the relative does not recover, the failure is attributed to God’s inscrutable will, insufficient faith, the presence of unconfessed sin, or the comforting notion that God had “a better plan.” This asymmetric attribution pattern guarantees that prayer will appear effective to the believer regardless of actual outcomes. Every recovery is a hit; every death is explained away. The hypothesis is confirmed by its successes and insulated from its failures.¹⁴¹⁵

The Base Rate Problem

A related epistemological challenge is the base rate problem. Many medical conditions have high spontaneous recovery rates. The common cold resolves on its own. Many cancers respond to treatment. Post-surgical patients usually heal. When prayer is offered alongside medical care—as it almost invariably is—any positive outcome is overdetermined: it would have occurred with or without prayer.¹⁸

To demonstrate that prayer adds something beyond natural recovery and medical treatment, one would need to show that prayed-for patients recover at rates exceeding the base rate. This is precisely what the clinical trials attempted to measure—and consistently failed to find.⁵¹⁰ The base rate problem explains why anecdotal evidence for prayer is so abundant and so misleading: in a world where most people recover from most illnesses, prayer will appear to “work” most of the time simply because recovery is the default outcome.

Theological Implications of Null Results

The consistent failure of prayer to produce measurable effects under controlled conditions presents a genuine theological dilemma, even if many believers regard the studies as irrelevant to their faith.

If God exists and answers prayer, then controlled studies should, in principle, detect the effect—unless one adopts the position that God systematically refuses to act when being observed, which attributes to God a form of behavior indistinguishable from nonexistence (at least empirically). David Hume anticipated this line of reasoning in his discussion of miracles, arguing that testimony of the miraculous is never sufficient to establish a miracle because the probability of human error or deception always exceeds the probability that the laws of nature were actually violated.²¹

Alternatively, if prayer works but only in ways that are statistically undetectable, then its effects are so small or so inconsistent as to be practically indistinguishable from no effect at all. A God who answers prayer at a rate indistinguishable from chance raises the question of what “answering prayer” means in practice.¹⁴

Some theologians have embraced a third option: that prayer genuinely does not produce physical effects on the external world, and that this is theologically acceptable because prayer’s value lies elsewhere—in its capacity to foster spiritual growth, sustain hope, build community, and express dependence on God. This position, while coherent, requires abandoning or radically reinterpreting centuries of Christian teaching about petitionary prayer, including biblical passages that appear to promise concrete answers (e.g., Matthew 7:7–8, James 5:15, Mark 11:24).

Faith Healing and Medical Evidence

Claims and Investigations

Faith healing—the claim that prayer, the laying on of hands, or other spiritual practices can cure disease without medical intervention—represents the most dramatic assertion of prayer’s empirical efficacy. Such claims have been a feature of religious movements from antiquity to the present, including Pentecostal Christianity, Christian Science, and various New Age traditions.

Systematic investigation of faith healing claims has consistently failed to produce verified cases of miraculous cure. The Lourdes Medical Bureau, established by the Catholic Church in 1883 to evaluate claims of miraculous healing at the Lourdes shrine, has recognized only 70 miracles in over 140 years despite examining thousands of cases—a rate so low that it is easily accounted for by spontaneous remission, misdiagnosis, and psychosomatic factors.¹⁴ No faith healing claim has ever been validated under controlled conditions in which the diagnosis was confirmed before the alleged healing and the cure was verified by independent medical examination afterward.

High-profile faith healers such as Peter Popoff and Benny Hinn have been repeatedly exposed using techniques familiar to stage magicians: cold reading, planted audience members, concealed radio receivers, and the selective presentation of testimonials from those who feel they have been healed while ignoring the vast majority who have not.¹⁴

Medical Neglect and Harm

The most consequential real-world impact of belief in faith healing is the phenomenon of religion-motivated medical neglect—cases in which parents or guardians withhold conventional medical treatment from children in favor of prayer alone.

A landmark 1998 study by Seth Asser and Rita Swan in Pediatrics documented 172 child fatalities in the United States between 1975 and 1995 in families that withheld medical care on religious grounds. Of these, 140 cases involved conditions for which survival rates with medical treatment would have exceeded 90%. Many involved treatable conditions such as bacterial meningitis, pneumonia, diabetic ketoacidosis, and bowel obstructions.¹²

The children who died belonged disproportionately to denominations that teach exclusive reliance on faith healing, including the Church of Christ, Scientist; the Faith Assembly; the Followers of Christ; and the End Time Ministries. Several of these cases resulted in criminal prosecutions, though religious exemptions in many state child abuse and neglect statutes long shielded parents from legal accountability.¹²¹³

Swan, herself a former Christian Scientist whose child died of bacterial meningitis after the family relied on faith healing, became a leading advocate for the removal of religious exemptions from child protection laws. She documented how such exemptions created a two-tiered system of child welfare in which children in faith-healing families received less legal protection than their peers.¹³

These cases represent the starkest possible test of the faith healing hypothesis: when prayer is the sole intervention, outcomes are dramatically worse than when conventional medicine is used. The empirical record of faith healing applied without medical backup is not ambiguous—it shows clear, measurable harm.¹²

Prayer and Subjective Well-Being

While the evidence for intercessory prayer’s efficacy on third-party health outcomes is negative, a separate body of research has investigated whether personal prayer and spiritual practice benefit the practitioner. This research generally finds modest positive associations between religious practice and various measures of well-being, including lower rates of depression, greater reported life satisfaction, and stronger social support networks.¹⁶¹⁹

However, these associations are confounded by numerous variables: people who pray regularly also tend to belong to supportive communities, abstain from substance abuse, maintain stable relationships, and hold optimistic worldviews. Disentangling the specific contribution of prayer from these correlated lifestyle factors has proven difficult, and the effects that remain after controlling for confounds are modest.¹⁸

The psychological benefits of prayer—to the extent they exist—are readily explicable without invoking supernatural mechanisms. Meditation, journaling, and secular mindfulness practices produce comparable effects through similar pathways: focused attention, emotional regulation, cognitive reappraisal, and the placebo response. The question of whether prayer “works” for the pray-er is thus not a question about divine intervention but about the well-known psychological effects of contemplative practice.¹⁶

Ethical Considerations in Prayer Research

The ethics of prayer research have been debated from multiple perspectives. Richard Sloan has argued that physicians who recommend prayer to patients are overstepping the bounds of medical practice and imposing their religious values in a clinical setting, potentially alienating patients of different faiths or no faith.⁷¹⁸

Others have raised concerns about the ethical implications of the STEP study’s findings: if knowledge of being prayed for can produce a nocebo effect—as the data suggested—then informing patients that they are being prayed for could constitute a form of iatrogenic harm. The STEP study’s Group 3 results, in which patients who knew they were receiving prayer had a higher complication rate, suggest that the interaction between prayer, expectation, and anxiety is more complex and potentially more harmful than commonly assumed.⁴

There is also the question of resource allocation. The STEP study cost $2.4 million. Some have argued that spending research funds on hypotheses with no plausible mechanism of action—and no track record of positive results in prior studies—diverts resources from more promising lines of inquiry.¹⁷ Others counter that investigating widely held beliefs, even when they appear unlikely, is a legitimate function of science, particularly when those beliefs influence medical decision-making.

Conclusion

The empirical evidence on intercessory prayer is as clear as clinical research on any topic can be: across multiple well-designed studies involving thousands of patients, intercessory prayer produces no measurable improvement in health outcomes beyond what would be expected by chance.⁴⁵¹⁰ The largest and most rigorous study, STEP, found not merely a null effect but a possible adverse effect for patients who knew they were being prayed for.⁴

Theological responses to these findings range from the claim that God refuses to be tested to the argument that prayer’s purpose is subjective transformation rather than physical intervention. These responses are internally coherent within their respective theological frameworks, but they share a common feature: they render the prayer hypothesis unfalsifiable, placing it beyond the reach of empirical investigation. Whether this represents a legitimate boundary between science and theology or an ad hoc retreat from a disconfirmed hypothesis depends largely on one’s prior commitments.

The practical consequences of belief in prayer’s physical efficacy are most visible in faith healing contexts, where the substitution of prayer for medical care has resulted in documented, preventable deaths—particularly among children.¹² Whatever one’s theological convictions about prayer’s unseen effects, its visible effects when used as a substitute for medicine are unambiguously harmful.

The question of prayer and empirical evidence ultimately illustrates a broader tension in the relationship between religious belief and scientific inquiry. Science asks what can be measured, tested, and replicated. Religion asks what can be trusted, hoped for, and believed. The clinical trial literature on prayer suggests that these two modes of inquiry may be genuinely incommensurable—not because science has failed to ask the question properly, but because the question itself may not be the kind that empirical methods can answer.⁸¹⁶