Argument

Argument from Miracles

Intro

If even one miracle has really happened, atheism is in trouble. Not slightly. Fundamentally. A miracle is an event nature cannot produce on its own, so if a miracle has occurred, there must be something beyond nature that can act inside nature. That something is what people have always called God.

The argument here does not assume miracles. It looks at the historical record and asks whether any specific cases are best explained by divine action. The strongest case by far is the resurrection of Jesus, which the page treats as its main example. But there are many smaller cases too. Craig Keener's two-volume Miracles (2011) catalogs thousands of medically-documented, witness-attested healings from the modern world, from every continent, including events in the last fifty years that doctors cannot explain.

The objection most people raise comes from David Hume in 1748: miracles are violations of natural law, natural law is established by uniform experience, so miracles cannot happen. The page argues Hume's reasoning is question-begging. It rules out miracles by assuming in advance what it was supposed to prove, that experience really is uniform. Even philosophers who do not believe in God (John Earman is the famous example) have shown Hume's argument fails on basic probability theory.

If miracles happen, nature is not a closed box. And if even one miracle has actually happened, then God exists and acts in history. That is what makes the argument apologetically explosive.

In full

A historical-evidential argument: events that exceed the operations of nature, miracles, when historically well-attested, are best explained by divine intervention. The strongest miracle-claim is the resurrection of Jesus Christ; numerous historical, hagiographic, and contemporary miracle-claims (Keener's two-volume catalog) add cumulative weight. If miracles occur, the natural order is not closed, and a supernatural agent must exist who can act within it. The argument's central battlefield is the Hume critique, and as Earman (Hume's Abject Failure, 2000) shows, Hume's a priori dismissal collapses on careful inspection. Structured as debate prep: each premise carries a second-order positive case, anticipated objections in the opponent's voice, numbered rebuttals naming failure-modes, a live-cite kit, and tactical notes.

Argument structure

Form

Combination of (a) abductive / inference-to-best-explanation, given the historical data, the miraculous-occurrence hypothesis is the best explanation; and (b) Bayesian, the prior probability of miracles is conditional on the prior probability of theism; given even a moderate prior for theism, the posterior probability of historically-attested miracles is substantial. Swinburne (The Resurrection of God Incarnate, 2003) develops the Bayesian formalization. The argument is the historical complement to the cosmological / teleological / moral arguments, God is not only the necessary condition of the universe's existence, order, and morality, but also a personal agent who acts in history.

P1, Miracles, if they occur, require a non-natural cause

Affirmative case (second-order arguments)

1. Definitional. A miracle is, by the standard Augustinian / Aquinian / Swinburnean definition, an event "that is not produced by the operations of nature" (Aquinas, SCG III.101) or, more carefully, an event "of religious significance brought about by a god" (Swinburne, The Concept of Miracle, 1970). The claim that miracles, if they occur, require a non-natural cause is analytic, it follows from how the term is defined. The substantive question is not "does miracle-by-definition require God?" but "do any actual events meet the definition?"
2. The principle is not God-of-the-gaps. The reasoning is not "we don't currently understand X, therefore God." It is: "X exceeds what nature can in principle do (raise the dead bodily, instantaneously regenerate organic damage, restore a 4-day-decomposed corpse), therefore the cause is by definition non-natural." The relevant test is in-principle natural impossibility, not current scientific puzzlement.
3. The methodological-naturalism / metaphysical-naturalism distinction matters here. Plantinga (Where the Conflict Really Lies, 2011) distinguishes the methodological assumption that "in scientific investigation, we look for natural causes" from the metaphysical claim that "only natural causes exist." The first is a useful research-program rule; the second is a dogma without independent justification. Premise 1 is consistent with methodological naturalism in science but rejects metaphysical naturalism's a priori closure of the natural order.

Anticipated objections

1. "Miracles by definition violate natural law; therefore, by definition, they don't happen." Hume's circular move; many contemporary repetitions.
2. "Any apparent miracle is just an event we don't yet have the natural explanation for." Carl Sagan-style methodological skepticism.
3. "The 'non-natural cause' you posit is unfalsifiable, which makes the argument unscientific."

Rebuttals

1. The Humean move is question-begging, failure-mode is defining miracles as impossible to avoid evidence. "Miracles violate natural law; natural law is established by uniform experience; therefore miracles cannot happen." But uniform-experience-establishing-natural-law is itself the question at issue: the miracle-claim is a claim of non-uniform experience. Hume's argument rules out miracles by defining the inquiry in a way that excludes the data. Earman (Hume's Abject Failure, 2000), himself an agnostic, demonstrates that Hume's argument is logically defective on multiple grounds: it confuses frequency with probability, treats base-rate as posterior probability, and assumes the very closure-of-nature it is supposed to establish. Earman's conclusion: Hume's Of Miracles is the worst essay in modern philosophy of religion.
2. The "we don't yet have the natural explanation" deflection collapses on the in-principle test. Some events are currently unexplained (frontier of science). But miracle-claims target events that are in principle naturally impossible, bodily resurrection of a 3-days-dead corpse, restoring an amputated limb instantly, calming a storm by command. These are not gap-claims; they are claims about events beyond what nature can in principle do (per the laws of physics, biology, chemistry as we currently and even prospectively understand them). The naturalistic-fix-pending move requires hand-waving at unspecified future science to avoid present evidence.
3. The unfalsifiability charge cuts the wrong way, Christianity is uniquely falsifiable on its central miracle-claim. Paul stakes Christianity on a single historically-falsifiable claim: "if Christ has not been raised, your faith is in vain" (1 Cor 15:14). Show me the body, and Christianity is falsified. The opponent's claim, "miracles are impossible no matter what evidence is presented", is the unfalsifiable position, not Christianity's. Failure-mode: the objection inverts which side is falsification-resistant.

Live-cite kit

• Scripture: John 14:11 (Jesus appeals to His miracles as evidential); 1 Cor 15:14 (the falsifiability stake)
• Scholarly: Swinburne (The Concept of Miracle, 1970, definitional clarity); Plantinga (Where the Conflict Really Lies, 2011, methodological/metaphysical naturalism distinction); Earman (Hume's Abject Failure, 2000, Hume rebuttal); Aquinas SCG III.101
• Aphorism: "If by 'miracle' you mean 'thing nature cannot do,' then yes, that's what we mean. The question is whether any have happened."

Tactical notes

• Force the opponent to commit on the methodological-vs-metaphysical-naturalism distinction early. Most opponents conflate them; making the distinction explicit forces them off the closed-system dogma.
• Use Earman by name, the fact that an agnostic philosopher of science wrote the definitive Hume-on-miracles rebuttal lands hard.
• Don't defend every reported miracle on this premise, the premise is about the kind of cause a real miracle would require, not about which specific reports are credible. Defer the credibility battles to P2.

P2, At least some historical events are best explained as miraculous

Affirmative case (second-order arguments)

1. The resurrection of Jesus is the strongest single historical miracle-claim. Multiple independent attestation (Gospels + Pauline + creedal); early-creedal evidence (1 Cor 15:3-8 within ~5 years of crucifixion); empty tomb; multi-modal post-mortem appearances; transformation of disciples; conversion of skeptics (Paul, James). See Argument from the Resurrection for the full case. The naturalistic alternatives (hallucination, swoon, theft, legend) each fail to explain the convergent data.
2. Craig Keener's two-volume Miracles (2011) catalogs thousands of contemporary documented cases. Keener, Asbury professor, mainstream NT scholar, compiles ~1100 pages of documented contemporary miracle-claims, mostly from the global South (Africa, Asia, Latin America), many with medical documentation. His explicit purpose is to corrective the Western secular assumption that miracles do not happen in modern life. The volume is methodologically careful: distinguishes weakly-attested from medically-documented cases; separates eyewitness from secondhand reports.
3. Multiple lines of contemporary medical-research evidence. Candy Gunther Brown's Testing Prayer (2012, Harvard University Press) examines healing prayer in randomized-controlled studies; the Brown-Mosteller studies on auditory and visual healings in Mozambique. Not all results are positive, but the existence of medically-documented unexplained recoveries following prayer is established in mainstream medical literature (cf. Lourdes Medical Bureau's century of cases).

Anticipated objections

1. "Hume's a priori, even strong testimony for miracles is always less probable than the testimony being mistaken or fraudulent."
2. "Other religions also claim miracles (Hindu, Islamic, Buddhist), does this not cancel out?"
3. "Naturalistic alternatives to the resurrection (hallucination, swoon, theft, legend) suffice."
4. "Keener's contemporary cases are unverified, anecdotal, or attributable to placebo / spontaneous remission / misdiagnosis."

Rebuttals

1. The Hume rebuttal, Bayesian formalization shows the inference can run. Earman (Hume's Abject Failure, 2000) shows Hume's argument fails on Bayesian grounds: if even a small prior for theism is granted (P(theism) > 0), then the prior on miracles is non-zero, and strong testimonial evidence can outweigh the Humean prior. Swinburne (The Resurrection of God Incarnate, 2003) provides explicit Bayesian formalization for the resurrection: even with conservative priors, the posterior probability of the resurrection given the evidence is substantial. The McGrew couple (Timothy and Lydia McGrew, "The Argument from Miracles" in Blackwell Companion to Natural Theology, 2009) provide further Bayesian rigor. Failure-mode: Hume confuses base-rate frequency with the operative posterior probability.
2. Not all miracle-claims are equally well-attested, the Christian resurrection is uniquely well-anchored. The principle "many religions claim miracles, so they cancel out" treats all miracle-claims as evidentially equivalent, which is false. Datable creedal material within 5 years (1 Cor 15:3-8); named eyewitnesses inviting verification (1 Cor 15:6); enemy and skeptic conversions (Paul, James); willingness-to-die under torture (apostolic martyrdom). No comparable miracle-claim from another tradition has this evidential apparatus. Press the opponent: "Name one other miracle-claim with comparable historical-critical anchoring." None exists. Other miracle-claims must be assessed individually on their evidence, and most fail the test that Christianity passes.
3. Naturalistic alternatives to the resurrection each fail multiple tests of explanatory adequacy. Hallucination is private/brief/confirmatory; the appearances were shared/prolonged/contradictory. Swoon is medically impossible per Roman crucifixion + spear-thrust (Edwards et al., JAMA 1986). Theft fails on disciples' willingness-to-die for what they would have known to be a lie. Legend fails on the 1 Cor 15:3-8 dating (within 5 years, too compressed for legendary development per Sherwin-White's Roman-historiography norms). See Argument from the Resurrection master objections.
4. Keener is methodologically careful, distinguishes well-documented from anecdotal cases. Keener's Miracles (2 vols, 2011) explicitly grades cases by evidential strength. Some are weakly attested (single witness, no medical record); others are well-documented (medical records before/after, multiple independent witnesses, medically-unexplained reversals). The objection conflates Keener's bulk with his best cases. Plus: even granting that some Keener cases are dubious, it takes only one well-documented case to refute the closed-system claim. The Lourdes Medical Bureau alone, operating since 1883 with strict naturalistic-elimination protocols, has acknowledged ~70 cases as medically inexplicable. Failure-mode: the objection demands no false positives but ignores the asymmetry (one well-attested miracle suffices).

Live-cite kit

• Scripture: Acts 2:22 ("Jesus the Nazarene, a man attested to you by God with miracles and wonders and signs"); Acts 13:32-33; Heb 2:3-4 ("God bearing witness to the gospel both by signs and wonders and by various miracles"); John 20:30-31 (the sēmeia, signs as evangelistic evidence); 1 Cor 15:3-8
• Scholarly: Craig Keener (Miracles, 2 vols, Baker Academic, 2011); Candy Gunther Brown (Testing Prayer, Harvard, 2012); Swinburne (The Resurrection of God Incarnate, 2003); McGrew & McGrew (Blackwell Companion to Natural Theology, 2009); Wright (The Resurrection of the Son of God, 2003); Habermas (The Resurrection of Jesus, 2024); Licona (The Resurrection of Jesus, 2010)
• Aphorism: "Show me one verified case nature cannot account for, and the closed-system claim is dead. Keener gives you 1100 pages of candidates."

Tactical notes

• Lead with the resurrection (the load-bearing case) and reserve Keener for the cumulative-weight tail. The resurrection alone, if granted, runs the argument; Keener piles on.
• If the opponent retreats to "all miracle-claims are equally weak," call the Lourdes Medical Bureau specifically, it operates under strict naturalistic-elimination protocols administered by doctors (often non-Catholic). Concrete protocol-based cases land harder than vague "many people claim miracles."
• Don't defend every Keener case, concede that some are weakly attested and double down on the well-documented cases plus the resurrection.

P3, Therefore a non-natural agent capable of acting in history exists

Affirmative case (second-order arguments)

1. Modus ponens from P1 + P2. If miracles require a non-natural cause (P1), and miracles have occurred (P2), then a non-natural cause-of-miracles exists. The inference is straightforward.
2. The cause has the personal-agency profile of theism, not deism or pantheism. A miracle is a willed intervention into the natural order with a specific historical and theological purpose, which requires intentional agency, not impersonal force. The cause must be a personal agent capable of overriding the laws of nature for specific ends, the classical-theistic God.
3. The cumulative-case complement. This argument intersects with the cosmological (necessary cause-of-existence), teleological (designer-of-fine-tuning), and moral (moral-lawgiver) arguments. Each argument identifies a different feature of God, the miracle argument identifies God-as-historical-agent. The cumulative case sees the same God across all four lines of inquiry, satisfying the multi-criterion comparative analysis in Christian God is the Only True God.

Anticipated objections

1. "Even granting miracles, why must the cause be God? Maybe it's an alien, an unknown force, a 'cosmic mind' rather than the Christian God."
2. "This still doesn't establish Christianity, Hindu miracles would establish a Hindu God."

Rebuttals

1. The "alien / unknown force" deflation runs into the explanatory-economy demand and the religious-context of the miracles. Miracles in the Christian record occur in religious context with specific theological purpose (vindicating Christ's claims, attesting apostolic teaching, sanctifying the saints). An impersonal "unknown force" doesn't explain the religious-purposive character of the events. Aliens require a multi-step ad-hoc rescue (alien existence + alien capability of mimicking divine intervention + alien interest in mimicking biblical theology) and have far worse explanatory economy than direct theistic intervention. Failure-mode: the deflation multiplies entities to avoid the obvious inference.
2. The Hindu-miracle parry runs into the comparative-evidential test of P2. Other religions' miracle-claims must be assessed individually on their evidence; most lack the historical-critical anchoring Christianity has. The cumulative-case comparison (Christian God is the Only True God) addresses this systematically. The objection assumes evidential-equivalence across religions, which is empirically false.

Live-cite kit

• Scripture: Heb 2:3-4 (God's miracle-attestation of the gospel); Acts 14:3
• Scholarly: Swinburne (The Existence of God, 2nd ed. 2004, cumulative-case Bayesian); Craig (Reasonable Faith, 2008, cumulative-case framing); Keener (Miracles, religious-context attestation)
• Aphorism: "Miracles in religious context attest the religion. Hostile witnesses know this, that's why ancient Jews and Romans had to attribute Jesus's miracles to demonic power rather than deny them (Mt 12:24)."

Tactical notes

• Press: "Granted miracles happen, what kind of agent best explains them?", and let the opponent name the alternatives. Most cannot defend a non-theistic agent that fits the data.
• The Christian-particularization (vs. Hindu, Islamic, etc.) belongs to the comparative-religion argument Christian God is the Only True God, defer there if the opponent presses.

Conclusion

God exists and acts in history. The historical record contains events that cannot be explained by natural processes alone. The Christian gospel rests on the historical reality of the resurrection of Jesus; if that event occurred (and the evidence supports it), then God exists and has acted decisively in history. The argument from miracles is the historical complement to the cosmological / teleological / moral arguments, God is not only the necessary condition of the universe's existence, order, and morality, but also a personal agent who acts in history.

Master objections to the whole argument

1. "The principle of analogy (Troeltsch): historical reasoning must analogize to present experience; events without analogue cannot be historically established." Reply: too restrictive, taken seriously, the principle rules out unique historical events generally (the Big Bang, the origin of life, the rise of consciousness, any first-of-its-kind event). Methodologically a-historical. Plus: Keener's catalog establishes contemporary miracle-claims, providing the very analogue Troeltsch demands.
2. "Closed-system naturalism is a methodological assumption of modern science." Reply: methodological naturalism is consistent with this argument (per Plantinga's distinction in P1 rebuttal 1). The argument targets metaphysical naturalism, which is a dogma without independent justification. Conflating methodological-and-metaphysical naturalism is the standard secular sleight.
3. "Bayesian formalization just hides the prior assumption of theism." Reply: any formalization requires priors; the question is whether the priors are reasonable. Swinburne and the McGrews defend modest theistic priors on independent grounds (cosmological arguments, simplicity considerations); the priors are not arbitrary. And even with low priors, the resurrection evidence is strong enough to drive substantial posterior probability, see Swinburne's calculation in The Resurrection of God Incarnate (2003).
4. "Cessationism, even granting biblical miracles, they ceased after the apostolic age." Reply: this is an intra-Christian debate (cessationist vs. continuationist), not a theist-vs-naturalist debate. Both cessationists and continuationists affirm the historical reality of biblical miracles; the question is whether they continue. The argument runs on biblical miracles alone; Keener's contemporary cases are bonus, not load-bearing.

Tactical opening / closing

Opening line: "Either nature's order has been intervened upon, or it hasn't. Christianity says it has. Let's look at the strongest case, the resurrection, and Keener's two volumes of contemporary documented claims. Then you tell me whether 'closed-system naturalism' is a finding or a dogma."

Closing landing strip: "If even one well-documented miracle has occurred, naturalism's central claim, that the natural order is closed, is false. The argument doesn't compel; it provides cumulative warrant. The relevant question is whether you'll examine the evidence on its merits, or rule it out a priori the way Hume did."

Connection to Scripture

• Acts 13:32-33, Pauline preaching of the resurrection as God's fulfilling intervention in history
• John 20:30-31, sēmeia (signs) function evangelistically
• John 14:11, Jesus appeals to His miracles as evidential
• Acts 2:22, Peter at Pentecost: "a man attested to you by God with miracles and wonders and signs which God performed through Him in your midst, just as you yourselves know", apostolic argument from miracles built into the earliest Christian preaching
• Hebrews 2:3-4, God bearing witness to the gospel "both by signs and wonders and by various miracles"
• 1 Corinthians 15:3-8, the early creed of the resurrection appearances; foundational historical claim

Patristic / scholarly note

Classical / patristic / medieval:

• Irenaeus (Against Heresies 2.32.4-5), cites contemporary (2nd c.) miracles as confirming apostolic gospel
• Augustine (City of God 22.8-10), compiles catalog of contemporary miracles, refuting cessation-with-the-apostles claim
• Gregory of Tours (De Gloria Martyrum, c. 590), major medieval miracle-catalog source
• Aquinas (ST III, q. 43; SCG III.101), theological treatment of miracle as definitionally non-natural

Reformation:

• Reformers (Luther, Calvin) largely cessationist about charismatic miracles (the giving of new revelation) but affirm historical reality of biblical miracles
• The cessationist-continuationist debate is intra-Christian, not theist-vs-naturalist

Modern:

• David Hume (Enquiry, 1748, §10), locus classicus for the modern attack
• William Paley (Evidences of Christianity, 1794), classical Anglican defense
• C. S. Lewis (Miracles, 1947), popular philosophical defense
• Norman Geisler (Miracles and the Modern Mind, 1992)
• Richard Swinburne (The Concept of Miracle, 1970; The Resurrection of God Incarnate, 2003), leading contemporary Bayesian-historical defender
• William Lane Craig (Reasonable Faith, 1984; rev. 2008; Assessing the New Testament Evidence, 1989)
• N. T. Wright (The Resurrection of the Son of God, 2003)
• Gary Habermas (The Risen Jesus & Future Hope, 2003; The Resurrection of Jesus, 2024)
• Mike Licona (The Resurrection of Jesus, 2010)
• Craig Keener (Miracles, 2 vols, 2011), most extensive recent compilation of contemporary miracle-claims
• Candy Gunther Brown (Testing Prayer, 2012), medical evidence in randomized-controlled studies
• Timothy and Lydia McGrew ("The Argument from Miracles" in Blackwell Companion to Natural Theology, 2009), Bayesian rigor
• John Earman (Hume's Abject Failure, 2000), definitive Hume rebuttal

See also

• Argument from the Resurrection, the prototype miracle and primary evidential focus
• Resurrection-Centric Growth Argument, apostolic-martyrdom companion
• Liar Lunatic or Lord, Christological dimension; Jesus's miracle-claims and self-claims fit together
• Argument from Prophecy Fulfillment, parallel evidential argument
• Fine-Tuning Argument, general providential intervention in cosmic constants; miracles are specific providential interventions
• Christian God is the Only True God, cumulative-case where miracles serve as P4 evidence
• Naturalism, the rebuttal target
• Richard Swinburne (entity, pending)
• Craig Keener (entity, pending)
• Hume on Miracles (concept, pending)
• Arguments, master index