GLP-1 Drugs and Alcohol: What the Trials Actually Show

The reports came from patients first: people who started semaglutide for weight or for diabetes and noticed, unprompted, that they had stopped wanting to drink. That anecdote has since produced two large observational studies, three randomized trials, and a great deal of coverage that treats all five as the same kind of evidence. They are not. This article separates the randomized trials from the database work, and says which endpoints each trial actually hit — including the one that missed.

Three randomized trials, 206 people between them

That is the entire randomized evidence base as of August 2026. Two hundred and six people, three single-center or academic-center trials, none longer than 26 weeks, none large enough to support a regulatory claim. Every one of them describes itself as a phase 2 study and every one of them ends with a call for larger trials. That framing is not a hedge added here — it is the authors' own.

The first trial, and the narrowest. Hendershot and colleagues randomized 48 people out of 504 assessed, at a single US academic medical center, to nine weeks of subcutaneous semaglutide (escalating to a maximum of 1.0 mg for one week) or placebo. Crucially, participants were non-treatment-seeking — they had alcohol use disorder but were not trying to quit — and the primary outcome was not real-world drinking at all but a laboratory alcohol self-administration task. On that primary outcome the drug worked: "Low-dose semaglutide reduced the amount of alcohol consumed during a posttreatment laboratory self-administration task, with evidence of medium to large effect sizes for grams of alcohol consumed (β, -0.48; 95% CI, -0.85 to -0.11; P = .01)¹."

On real-world drinking, the results split. "Semaglutide treatment did not affect average drinks per calendar day or number of drinking days," but it "significantly reduced drinks per drinking day (β, -0.41; 95% CI, -0.73 to -0.09; P = .04) and weekly alcohol craving (β, -0.39; 95% CI, -0.73 to -0.06; P = .01)¹." In plain terms: participants drank on just as many days, but drank less on the days they drank. The authors' own conclusion is deliberately modest — "initial prospective evidence... justifying larger clinical trials."

The largest and longest trial. Klausen and colleagues in Denmark randomized 108 treatment-seeking participants with moderate to severe alcohol use disorder and comorbid obesity to 26 weeks of semaglutide 2.4 mg or placebo, with both arms also receiving standard cognitive behavioral therapy. The primary endpoint — reduction in heavy drinking days at 26 weeks — was met: semaglutide produced a fall of 41.1 percentage points from baseline (95% CI -48.7 to -33.5) against 26.4 points on placebo, "estimated treatment difference -13.7 percentage points, -22.0 to -5.4; p=0.0015²."

Read those three numbers carefully, because the headline version drops one of them. The placebo arm fell 26.4 points on its own. Roughly two thirds of the total improvement in the semaglutide arm is accounted for by everything that was not the drug — the therapy, the trial structure, regression to the mean, the effect of being watched. The drug's own contribution is the 13.7-point difference, which is real, statistically robust, and about a third the size of the number most coverage quotes. The trial also required comorbid obesity for entry, which leaves open whether the same effect appears in people with alcohol use disorder and a normal BMI.

The trial that missed its primary endpoint. Schacht and colleagues randomized 50 treatment-seeking adults with moderate to severe alcohol use disorder to eight weeks of oral semaglutide or placebo. The primary outcome was laboratory-based alcohol cue-elicited craving at week 6, and it failed: "Semaglutide did not significantly reduce laboratory-assessed craving or drinks per day compared with placebo³."

What the same trial did find, on preregistered secondary outcomes, was a significant reduction in heavy drinking days (b = -0.580, 95% CI -1.012 to -0.148), in drinks per drinking day, in naturalistic (self-reported, real-world) craving, in alcohol-related negative consequences, and in cannabis use days³. That is a genuinely mixed result, and how it gets reported matters: a trial whose primary endpoint misses and whose secondaries hit is hypothesis-generating, not confirmatory, no matter how many secondaries hit. The authors' own summary — "These findings confirm previous results among non-treatment seekers with less severe AUD and suggest that continued development of semaglutide for AUD is warranted" — is a fair reading, but "the primary endpoint was not met" belongs in every summary of this trial and is usually missing from them.

Note also that Schacht's trial used the ORAL formulation, not the injection — our article on oral semaglutide covers how that route differs, which matters here because the two positive trials both used injections.

The observational studies that started all this

The randomized trials exist because two very large database studies pointed at something first. Both are worth reading, and neither is a trial.

The Swedish study is the more rigorous design. Lähteenvuo and colleagues took all 227,866 residents aged 16 to 64 with a diagnosis of alcohol use disorder from national registers spanning 2006 to 2023 and ran a within-individual Cox model — comparing each person's risk during periods when they were taking a GLP-1 drug against their own risk when they were not, which removes the confounding by fixed personal characteristics that wrecks most drug-versus-nondrug comparisons. Semaglutide use was associated with the lowest risk of hospitalization for alcohol use disorder (adjusted hazard ratio 0.64; 95% CI 0.50 to 0.83), liraglutide the second lowest (0.72; 0.57 to 0.92), while "use of any AUD medication was associated with a modestly decreased risk (aHR, 0.98; 95% CI, 0.96-1.00)⁴."

Two limits sit inside those numbers. Only 4,321 of the 227,866 people ever used semaglutide, so the entire estimate rests on a sliver of the cohort. And a within-individual design still cannot control for why someone started a GLP-1 in the first place — a period of being motivated enough to seek and take a weekly injection is not a random period of anyone's life. The authors say so themselves, ending with "clinical trials are urgently needed to confirm these findings." Our liraglutide review covers that older drug's own approval history, which is relevant here because it is the only other agent in this cohort with a comparable signal.

The US study is a retrospective analysis of electronic health records: 83,825 patients with obesity, comparing semaglutide against other anti-obesity medications, finding "a 50%-56% lower risk for both the incidence and recurrence of alcohol use disorder for a 12-month follow-up period," with similar findings replicated in 598,803 patients with type 2 diabetes⁵.

That paper carries a published correction, and what it corrected is directly relevant to how it gets cited. The original text claimed semaglutide was associated with a significantly lower risk of INCIDENT alcohol use disorder compared with naltrexone or topiramate — two drugs actually approved or used for alcohol use disorder. The correction changes that specific claim to RECURRENT diagnosis, and separately corrects the stated follow-up window from 8 months to 12⁶. If you have seen this study cited as showing semaglutide beats naltrexone at preventing alcohol use disorder from developing, that is the uncorrected version.

What the FDA labels say about alcohol: nothing

This is worth stating precisely because the absence is the point. Both current semaglutide labels were fetched and searched in full for this article. Wegovy's label contains the word "alcohol" in exactly two contexts: "alcohol swab" and "alcohol wipe" in the Instructions for Use, and "nonalcoholic steatohepatitis" in the description of its liver indication⁷. There is no alcohol-related indication, no warning, no precaution, no adverse reaction, and no drug-interaction entry. Ozempic's label is the same.

No GLP-1 drug is approved anywhere for alcohol use disorder. Any provider offering one for that purpose is prescribing off-label on the strength of three phase 2 trials totaling 206 people, and should say so.

What the evidence supports, tier by tier

The pattern across all three trials is more specific than "it reduces drinking," and it is consistent enough to be worth naming: what moves most reliably is the amount consumed per drinking occasion and the intensity of craving. What moves least reliably is whether people drink at all on a given day. Hendershot found drinking days unchanged; Schacht found drinks per day unchanged; both found drinks per drinking day reduced. If there is a real mechanism here, the current data are more consistent with it blunting how much you drink once you start than with it removing the impulse to start.

What this means if you are considering it

Nothing on this site's boards is sold for alcohol use disorder, and this article is not an argument that it should be. Three points are worth being direct about:

The trials used pharmaceutical-grade drug at label-defined dosages. Every figure above comes from a trial administering a branded or trial-supplied product under supervision. It says nothing about a compounded vial — our explainer on compounded semaglutide covers what that is and is not, and none of the alcohol evidence transfers to a product whose identity and concentration are not FDA-reviewed.

The side-effect profile does transfer. Whatever the alcohol signal turns out to be, someone taking semaglutide for it takes on the same gastrointestinal reactions, the same boxed thyroid warning, and the same monitoring requirements as anyone taking it for weight — our semaglutide dosing reference covers what the labels actually specify. Klausen's trial reported adverse events that "were transient, generally mild to moderate gastrointestinal effects, and occurred more frequently in the semaglutide group²."

Approved treatments for alcohol use disorder already exist, and the one comparison in this literature that put GLP-1 drugs against them is observational, not randomized. Nobody has run a head-to-head trial. For what providers charge for semaglutide itself, our semaglutide provider rankings carry the price research, and our tirzepatide board covers the other approved incretin — neither drug is sold, or should be sold, on an alcohol claim.

The bottom line

Three randomized trials totaling 206 participants have tested semaglutide for alcohol use disorder. Two met their primary endpoint and one did not. The largest showed a real but modest drug effect — 13.7 percentage points of heavy drinking days beyond a placebo arm that improved by 26.4 points on its own. Across all three, reductions in craving and in drinks per drinking day are more consistent than reductions in how often people drink. The much larger observational studies point the same direction, one of them with a published correction that narrows its most-quoted comparative claim. And no FDA label for any GLP-1 drug mentions alcohol at all.