Unanswered Prayer

In 1960, C. S. Lewis's wife Joy Davidman died of cancer after a period of remission during which Lewis and others had prayed fervently for her healing. His journal, published as A Grief Observed under a pseudonym, contains some of the starkest religious writing on divine silence produced in the twentieth century: 'Go to Him when your need is desperate, when all other help is vain, and what do you find? A door slammed in your face, and a sound of bolting and double bolting on the inside. After that, silence.' This was not written by a skeptic looking for ammunition but by perhaps the most influential Christian apologist of his era, at the height of his grief.

A generation later, philosopher Nicholas Wolterstorff wrote Lament for a Son after his 25-year-old son Eric died in a mountain-climbing accident. Wolterstorff does not stage a philosophical argument against theism; instead he sits with unresolved pain, refusing false comfort while continuing to address God directly in prayer even amid apparent silence. Together these two texts establish that the challenge of unanswered prayer is not a debater's talking point invented by critics, but a lived crisis faced by the tradition's most serious and articulate voices.

Bart Ehrman, in God's Problem, treats this pattern as symptomatic of a larger difficulty: if God is both willing and able to intervene, and prayer is the means by which believers request such intervention, then the routine absence of requested interventions is evidence, cumulatively, against the existence of such a God — or at least against the God described in the promissory language of much popular preaching and many scriptural texts taken at face value.

In 2006, a team led by cardiologist and mind-body medicine researcher Herbert Benson published results of the Study of the Therapeutic Effects of Intercessory Prayer, involving 1,802 patients undergoing coronary artery bypass surgery at six US hospitals. Patients were randomized into three groups: those prayed for and told they might or might not be prayed for, those not prayed for and told the same, and those prayed for and told with certainty that they would be. The study found no significant difference in complications between the first two groups, and, strikingly, a higher rate of complications (59% versus 52%) among patients who knew for certain they were being prayed for — a result the researchers attributed to possible performance anxiety ('what if I fail to get better despite people's prayers') rather than any negative spiritual effect.

The STEP study was far from the first attempt to test prayer empirically — earlier studies by Randolph Byrd (1988) and William Harris (1999) had reported modest positive effects, but these were smaller, less rigorously controlled, and have not been reliably replicated. A Cochrane systematic review of intercessory prayer studies through the mid-2000s concluded the evidence did not support a reliable positive effect on health outcomes. Taken together, the empirical literature gives no support to the idea that intercessory prayer functions as a reliable causal intervention detectable by clinical trial methodology.

Richard Sloan, a behavioral medicine researcher and author of Blind Faith, has argued the entire research program is conceptually confused from both a scientific and a theological standpoint: scientifically, because it is nearly impossible to control for all prayer (patients, families, and unaffiliated strangers may pray regardless of study design), and theologically, because it presupposes a model of God as a variable that should respond in statistically detectable, on-demand fashion to petition volume — a model most serious theologians explicitly reject.

It is worth noticing that the most theologically central prayer scene in the New Testament is itself a case of unanswered petition. In Gethsemane, facing crucifixion, Jesus prays: 'Father, if you are willing, take this cup from me; yet not my will, but yours, be done' (Luke 22:42). The cup was not taken away. Yet this prayer, not a story of granted requests, is presented across the Christian tradition as the paradigm of faithful prayer — a request submitted to, and ultimately subordinated to, a wisdom the one praying does not fully share.

The apostle Paul offers a similar pattern in 2 Corinthians 12:7-9, describing an unspecified 'thorn in the flesh' he asked God three times to remove, receiving instead the answer, 'My grace is sufficient for you, for my power is made perfect in weakness.' Denial, in Paul's telling, is not evidence of divine absence or indifference but is folded into a larger providential purpose Paul claims (rightly or wrongly) to have understood only in retrospect.

This matters for the argument from unanswered prayer because it shows that the New Testament's own central examples do not model prayer as a reliable transactional mechanism. Whatever popular preaching and simplified 'ask and you shall receive' readings of texts like Matthew 7:7 may suggest, the tradition's foundational narratives already build in the expectation that faithful prayer's 'answer' may not be the thing requested — which reframes, without eliminating, what a null result in a study like STEP should be expected to show.

A natural objection follows immediately: if any outcome, granted or denied, can be redescribed as 'God's wise answer,' has the claim that God answers prayer become compatible with literally any observation, and therefore empty of content? This echoes Antony Flew's classic 'falsification challenge' to religious language — the worry that theological claims quietly retreat from any possible disconfirmation until they say nothing at all.

The relational-model defender's best response is that this framework is not introduced ad hoc to dodge disconfirming evidence; it predates modern statistical prayer trials by millennia and is embedded in the tradition's foundational texts (Gethsemane, Paul's thorn). It is also, notably, structurally similar to how humans understand answers within any relationship involving a wiser or more knowledgeable party: a parent's 'no' to a child's request is a real, meaningful answer within a real relationship, and its occurrence is not evidence the parent does not exist or does not love the child, even though 'the parent sometimes says no' is compatible with a wide range of parental behavior.

Still, this analogy has limits. A parent's reasoning can eventually be explained to the child in ways that make the 'no' intelligible after the fact; God's reasons, in cases like a child's death or the failure of a desperately needed healing, often remain permanently opaque even to devoted believers like Lewis and Wolterstorff, who never claim to have received a satisfying explanation, only a chastened, ongoing relationship of trust maintained despite the absence of explanation.

Sorting through the empirical and theological material, a fairly clear picture emerges. The strongest form of the argument from unanswered prayer targets a genuinely popular and widespread theology: the idea, encouraged by decontextualized readings of verses like Matthew 7:7 ('ask and it will be given to you') and by much popular preaching, that sincere faith plus specific petition reliably yields the requested outcome. Against this folk theology, both personal testimony (Lewis, Wolterstorff) and empirical research (STEP) constitute real, substantial evidence of failure.

Against the tradition's more careful theological accounts — Gethsemane's model of submitted petition, Paul's thorn, Aquinas's and Calvin's treatments of prayer as participation in rather than override of providence — the argument's evidential force is considerably weaker, because these accounts never predicted the mechanistic reliability the argument tests for. This does not make the accounts unfalsifiable in a vicious sense, since they retain real content (a good and wise providence genuinely at work, even if opaque) that could in principle be undermined by other considerations (such as pervasive, gratuitous suffering with no discernible pattern of any redemptive good, the classical evidential problem of evil).

What remains, honestly, is a residue that no argument fully dissolves: even believers who accept the relational model and are philosophically satisfied that unanswered prayer does not refute theism can still find themselves, like Lewis at his wife's deathbed, experiencing God's apparent silence as an assault on faith rather than a philosophy problem to be solved. The distinction between an evidential and a pastoral problem is real and important, but it should not be used to dismiss the felt weight of the latter simply because the former has an adequate answer.