People Said GLP-1s Made Them Drink Less. Two Trials Put It to the Test
For years, people on GLP-1 medications reported wanting alcohol less — without trying to. Researchers have now tested it properly, and the finding is narrower and more interesting than the headlines suggest.
It started as an aside.
People who had gone on a GLP-1 medication for weight or blood sugar kept mentioning something nobody had asked them about: they didn't really want the glass of wine anymore. Not that they'd decided to cut back. It just stopped occurring to them.
For a while that was only a pile of anecdotes, which is the weakest kind of evidence there is — memorable, easy to repeat, and frequently wrong. Now it has been tested in actual trials, twice. The result is real, and it is narrower than most of the coverage suggests.
The short version
Two randomised trials have now found that semaglutide reduces how much people drink when they drink, along with how much they report wanting to. Neither found that it reliably reduces how often they drink.
That distinction is the entire story, and it is the first thing lost in a headline.
Key point
The effect showing up in trials is "smaller amount on a drinking day," not "stops drinking." Those are different findings with different implications, and only one of them is supported.
What the first trial did
A 2025 trial published in JAMA Psychiatry took 48 adults who drank at levels meeting the definition of alcohol use disorder — and who, notably, were not trying to cut down. That last detail matters. If you recruit people already attempting to quit, you can't tell whether the medication did anything or their own effort did.
They received low weekly doses of semaglutide, or a placebo, for nine weeks. Lower than the doses used for weight loss, deliberately, to keep it safe in a first study.
Two things were measured. One was what people reported week to week. The other was harder to argue with: participants came into a room, were given their own preferred drink, and could drink as much as they wanted over two hours while their breath alcohol was measured every half hour. No self-reporting, no memory involved.
In that room, the semaglutide group drank measurably less and reached a lower peak blood alcohol level.
In their ordinary lives, the picture was more mixed:
- Drinks on a day they drank — down
- Craving — down
- Heavy drinking days — down, increasingly so as the dose stepped up
- Number of days they drank at all — unchanged
- Average drinks per calendar day — no clear effect
They also lost about 5% of their body weight, which is what you'd expect at those doses. Side effects were the usual GLP-1 ones and were mostly mild.
What the second trial added
The 2025 study was small and short, and its authors said so plainly. A 2026 trial in The Lancet addressed most of that.
| 2025 trial | 2026 trial | |
|---|---|---|
| People | 48 | 108 |
| Length | 9 weeks | 26 weeks |
| Dose | Low (0.25–1.0 mg) | Full 2.4 mg |
| Who | Not seeking treatment | Seeking treatment, with obesity |
| Checked against | Self-report + lab session | Also blood markers of alcohol use |
Six months, a full dose, and — importantly — laboratory markers in blood that reflect heavy drinking, so the results don't rest on what people remember and choose to say. Everyone in both groups also received standard talking therapy.
The semaglutide group had significantly fewer heavy drinking days. Total consumption and standard drinking questionnaires pointed the same way, and so did the blood markers.
Two trials, different designs, different kinds of participants, same direction. That's a good deal more than an anecdote.
The trial that will actually settle it
Both studies so far were run at a single site, and 48 and 108 people is not enough to change how anyone is treated. The study designed to answer the question properly started recruiting on 28 July.
The Department of Veterans Affairs is running it, under the name CRAVE — Cessation or Reduction of Alcohol Consumption in Veterans. The scale is a different order of magnitude:
- More than 600 participants, aged 18 to 80, with moderate or severe alcohol use disorder
- 18 VA medical centres, spread from Seattle to Orlando
- Weekly injections of semaglutide or a placebo for around 24 weeks, then a safety follow-up
- Tracking not just drinking, but general health and quality of life
The choice of population isn't incidental. More than 400,000 veterans have been diagnosed with alcohol use disorder, and the VA is one of the few health systems in the country positioned to run a study this size across this many sites at once.
Note
Multiple sites matter more than the participant count alone. A result from one clinic can reflect that clinic — its staff, its local population, how it recruits. When the same effect turns up across eighteen centres in different parts of the country, it's much harder to explain away.
Results aren't expected until around 2028, with the study fully wrapping up in 2029. That is genuinely a long time to wait, and it is also how you find out whether something works rather than merely looks like it does.
Why would a stomach hormone do this?
This is the part worth understanding, because it explains why the effect is shaped the way it is.
GLP-1 is a hormone your gut releases when food arrives. It nudges the pancreas to release insulin, slows how fast the stomach empties, and sends a message upward that amounts to we're handling it, you can stop now. Medications like semaglutide are long-lasting copies of that message.
Here's the part people usually aren't told: that message doesn't only reach the parts of the brain that handle digestion. It also reaches the circuitry that produces wanting — the pull toward something rewarding, the mental tug that makes you reach for it before you've consciously decided anything.
That circuitry is not food-specific. It is the same general machinery involved in wanting alcohol, and nicotine, and a number of other things. So a drug that turns down the volume on wanting food may be turning down the volume on wanting in general.
Which fits the results neatly. It would explain why the drug reduces how much you drink once you start — the pull to keep going fades sooner — while leaving whether you open the bottle at all largely alone. That decision is habit, routine, and social context, and a satiety signal has much less to say about it.
Note
The same 2025 trial noticed something similar with cigarettes. Among the 13 participants who smoked, those on semaglutide cut down more than those on placebo. Thirteen people is far too few to conclude anything — but it points the same way, and larger studies are looking at it.
What this does not mean
Important
Semaglutide is not an approved treatment for drinking, anywhere. The completed trials are early-stage — 48 people, then 108 — at single centres, and the large one that could change that has only just begun enrolling. Nothing here is a reason to start, stop, or change a medication, and none of it should be arranged around drinking.
A few other honest caveats:
"Less" is not "none." Both trials measured reductions in heavy drinking, not abstinence. For some people that is a meaningful health improvement. For others it is not the goal.
The two groups weren't the same. The 2025 participants were at moderate severity and weren't trying to cut down; the 2026 group was more severely affected, seeking treatment, living with obesity, and receiving therapy alongside the injection. That the results agree is encouraging, but they aren't measuring quite the same people in the same situation.
Medications for drinking already exist, and have for decades — and fewer than one in fifty people who could benefit ever receive one. If alcohol is a live question for you, that conversation with your doctor is available today and doesn't depend on any of this research panning out.
The bottom line
Something people noticed by accident turned out, on inspection, to be real — which is not how these stories usually end.
But the finding is specific. Two trials suggest semaglutide reduces how much people drink on the occasions they drink, and how strongly they want to, probably because the signal it imitates reaches the brain's wanting machinery and not only its hunger machinery. It does not appear to change how often people drink. It is not approved for this, the completed studies are small, and the 600-person trial that could actually change practice only started enrolling weeks ago.
If you're on a GLP-1 and you've noticed your own relationship with alcohol shifting, you're not imagining it, and you're in fairly large company. It's worth mentioning to your prescriber — as information about how you're responding, which is genuinely useful to them.
For more on how these medications change wanting rather than willpower, see what happens when the food noise stops. For the physical side, see fatigue and feeling cold and what the research says about muscle loss.
This article is general education, not medical advice. Talk to your healthcare provider about your own treatment.