Few arguments in the atheist’s toolkit are as famous — or as frequently misused — as Epicurus’ Problem of Evil. It looks airtight. It feels devastating. And precisely because it seems so powerful at first glance, a lot of young atheists reach for it like it’s a mallet that cracks open any theistic position on contact.
It isn’t. Used carelessly, it’s weaker than you think. Used well, it’s one of the most compelling arguments in the philosophy of religion. The difference is worth understanding before you deploy it in a conversation.
The Classic Formulation
The Argument is usually stated something like this:
- If an omnipotent and omnibenevolent God exists, then evil cannot exist.
- There is evil in the world.
- Therefore, an omnipotent and omnibenevolent God does not exist.
Clean. Logical. Three lines, and apparently you’ve disproved God. What could go wrong?
Quite a bit, actually.
Why the Logical Version Runs Into Trouble
The standard theistic response is the free will defence, most famously developed by Alvin Plantinga. The idea goes like this: a world in which beings genuinely choose between good and evil is more valuable than one in which they are programmed always to choose good. For free will to be meaningful, the possibility of choosing evil must be real. God permitting the possibility of evil is therefore compatible with being omnibenevolent, because the alternative — a universe of moral puppets — would be a worse world overall.
Notice what this does to premise 1. It’s no longer obviously true. The theist is now claiming that an omnibenevolent God might have excellent reasons to allow evil. Unless you can definitively rule out such reasons, you can’t conclude that evil and such a God are logically incompatible.
You can push back. You can argue that an omnipotent God should be able to create beings with free will who nonetheless always freely choose good (the so-called “Mackie response”). You can argue that God is at least partially culpable for creating beings He knew would choose evil. You can argue about what “free will” even means, whether it’s coherent, whether it requires the actual existence of evil or just its logical possibility.
But now you’re deep in the weeds of metaphysics, and the conversation has drifted far from the crisp three-step syllogism you started with. The theist doesn’t need to win this sub-debate — they only need to make it plausible enough that your original premise is no longer obviously true. And they can usually do that.
So the logical problem of evil — the version that tries to prove God’s existence is impossible — has largely been abandoned by contemporary philosophers of religion, even atheist ones. It’s not dead, but it’s not the knockout punch people imagine.
Where the Classic Version Actually Works
There’s still a legitimate place for the original Argument, but it’s narrower than people realise. It works best as a response to specific theistic claims — not as a general disproof of God.
Consider the popular line of thinking where God has “a plan” for every event: He helps someone find their car keys, guides a football team to victory, answers a prayer about a job interview. In that framework, God is depicted as routinely intervening in trivial matters. If He’s willing to nudge the outcome of a sporting event, why won’t He prevent a rape? Why won’t He stop a child from being abused?
Here, the Argument isn’t trying to prove God doesn’t exist. It’s trying to show that a particular conception of God — the micromanaging, intervention-happy God of popular piety — is incoherent. You can’t have it both ways: either God intervenes selectively (in which case His priorities are monstrous) or He doesn’t really intervene in the car-keys-and-football cases either.
This is a useful deployment. It targets a specific claim rather than settling the entire question.
The Argument Gets Stronger When Free Will Is Off the Table
The free will defence only applies to evils humans cause. It has nothing to say about natural evil — earthquakes, cancers, congenital disabilities, tsunamis, the five-year-old dying of leukaemia, the animal suffering that has gone on for hundreds of millions of years with no human involvement whatsoever.
God preventing a tectonic plate from slipping doesn’t impinge on anyone’s free will. God preventing a genetic mutation that causes a child to be born unable to breathe doesn’t rob anyone of moral agency. There’s no plausible free-will-based justification for these. The theist has to reach for other defences — “soul-making theodicy,” mysterious divine purposes, the fallenness of nature after original sin — and each of these comes with its own significant problems.
This is where the Argument starts to bite hard. And it’s here that the more sophisticated version of the Problem of Evil really lives.
The Evidential Problem of Evil
The strongest form of the Argument, developed by philosophers like William Rowe, abandons the claim of logical impossibility and argues something more modest but more devastating: the existence of certain kinds of suffering makes the existence of an omnibenevolent God improbable, not impossible.
It can be stated roughly like this:
- There exists suffering that does not contribute to any greater good — suffering God could have prevented without losing anything of value.
- An omnibenevolent God would prevent such suffering.
- Therefore, it is unlikely that an omnibenevolent God exists.
Notice how much harder it is to dismiss this. The free will defence doesn’t help, because the premise only requires the existence of some gratuitous suffering — suffering that doesn’t serve free will, soul-building, or any other purpose. Rowe’s famous example is a fawn dying slowly over several days after a forest fire caused by lightning. No human is involved. No human will ever know about it. No soul is built. The animal suffers and dies.
What greater good does that serve? The theist has to either:
- Claim we can’t know that no greater good is being served (the “sceptical theism” response — which quickly undermines our ability to make any moral judgments about anything).
- Propose a specific greater good (and then defend why an omnipotent being couldn’t have achieved it some other way).
- Deny the premise and argue that no suffering is truly gratuitous (which becomes increasingly strained as examples pile up).
None of these responses is knockdown. But crucially, none of them can be invoked the way “free will” is invoked against the logical version. The evidential version demands real engagement with the specifics, and the specifics are not on the theist’s side.
The Practical Takeaway
If you’re going to use the Problem of Evil, be clear about which version you’re using and why.
The logical version — “evil proves God is impossible” — will get you into a metaphysical debate about free will that you’re unlikely to win, and that the theist doesn’t need to win to defuse your Argument. Don’t lead with it.
The targeted version — “your claim that God intervenes in trivial matters is incoherent given what He allows to happen” — is useful for pushing back on specific theistic claims about providence and prayer.
The evidential version — “the specific pattern and scale of suffering we observe makes an omnibenevolent God unlikely” — is the philosophical heavy hitter. It survives the free will defence. It requires the theist to actually engage with the specifics of suffering rather than retreating to abstraction. And it doesn’t overreach: it argues for improbability, not impossibility, which is both more defensible and, in most contexts, all you actually need.
There are other routes to arguing against theism, too — the evidential weight of divine hiddenness, problems with specific conceptions of God, the incoherence of various divine attributes, the Argument from religious diversity, and so on. The Problem of Evil isn’t the only tool in the shed, and treating it as the universal answer is part of why it gets misused.
Used carefully, it remains one of the most serious challenges to theistic belief ever formulated. Used carelessly, it makes the person wielding it look like they haven’t thought it through. The difference matters.
