Craving, Not Willpower, Is What Actually Predicts Relapse

Leslie Jamison didn’t lose her drinking to one bad night. She lost it to a slow contraction — “not just the hours I spent drinking,” she writes in her 2018 memoir The Recovering, “but the hours I spent anticipating drinking, regretting drinking, apologizing for drinking, figuring out when and how to drink again.” Somewhere inside that loop, she landed on her own definition of the thing every person in early recovery gets warned about but rarely told how to name: craving becomes pathology, she decided, “when it becomes tyrannical enough to summon shame.”

Craving is not a footnote to relapse.

It’s the mechanism — and it predicts relapse better than almost anything else researchers can measure: not how long you used, not what you used, not your age, income, or family history. A 2022 meta-analysis in JAMA Psychiatry pooled 237 studies and more than 51,000 people and found that a one-point rise in craving more than tripled the odds of using again. That is not a soft correlation buried in a table. That is the strongest signal addiction treatment currently has, and most programs still talk about it like an afterthought to “triggers.”

There are two versions of it, and if you’ve lived this, you already know the difference even if no one ever handed you the vocabulary. One is a low hum that just sits in your body for weeks after you stop — no trigger, no obvious cause, just background static of wanting. Researchers call this withdrawal craving. The other arrives out of nowhere: you’re doing fine, stacking days, and then a smell, a street corner, a stranger’s laugh through a doorway lands on you, and for ten unbearable seconds your whole nervous system organizes itself around one thing. That’s cue-induced craving — a person, place, or object that used to mean “using” hijacking your attention — and of the two, it’s the more dangerous. The same JAMA analysis found cue-triggered spikes carried the strongest link to relapse of any indicator studied.

You don’t get to choose which one shows up, and you rarely get much warning.

A 2024 study that ran machine learning over Reddit’s stopdrinking forum analyzed tens of thousands of anonymized posts and found something that reads less like data and more like a diagnosis of modern recovery: craving talk peaks hardest in the first days after the last drink, decays over time, but never fully disappears — and the words people reach for most when describing it are “bored” and “stressed,” not “sick” or “using.” People aren’t narrating desire so much as narrating a mood with nowhere left to go.

Mark Gold, M.D., the addiction researcher behind a recent Psychology Today piece on this exact question, put it plainly: craving doesn’t sit politely alongside your decision-making. It edits it — reweighting what counts as a reward before you’ve consciously decided anything at all.

Treatment programs still spend more time teaching the vocabulary of “triggers” than the physiology of craving itself. That’s backward. If craving outpredicts everything else clinicians track, it deserves to be the main event of relapse prevention, not the warm-up act.

Jamison never resolved her drinking into a clean before-and-after. She just kept naming the wanting until it got smaller than the shame required to keep hiding it. Most days, that’s the whole job — not eliminating the craving, but outlasting the ten seconds it takes to pass.