GLP-1s and alcohol: three randomized trials, and what each one moved
Three randomized trials of semaglutide in alcohol use disorder have now been published, in April 2025, May 2026 and September 2026. All three reduced heavy drinking. None of them reduced average drinks per day, two of them missed their own primary outcome, and every one enrolled people with a diagnosed alcohol problem — which isn't who's buying a GLP-1 from a telehealth platform.
What this comes down to
- The 2025 trial randomized 48 non-treatment-seeking adults for nine weeks. It cut alcohol consumed in a laboratory task and cut craving, but did not change drinks per calendar day or number of drinking days.
- The 2026 trial randomized 108 treatment-seeking people with alcohol use disorder and obesity for 26 weeks alongside cognitive behavioral therapy, and cut heavy drinking days by 13.7 percentage points more than placebo.
- The September 2026 oral semaglutide trial missed its primary outcome — laboratory cue-elicited craving — and still cut heavy drinking days and drinks per drinking day.
- Eight randomized trials in total pair a GLP-1 with an alcohol outcome. Naltrexone, the comparison every clinician will reach for, returns 201 through the identical filter.
The three trials, in order
April 2025, 48 participants, nine weeks
A phase 2 double-blind parallel-arm trial at a US academic medical center randomized 48 non-treatment-seeking adults with alcohol use disorder to weekly semaglutide, escalated to 1.0 mg, or placebo[1]. The primary outcome was laboratory alcohol self-administration, and it moved: grams of alcohol consumed fell with a coefficient of -0.48 and peak breath alcohol concentration by -0.46, both with confidence intervals clear of zero.
What didn't move is just as informative. Semaglutide didn't affect average drinks per calendar day or the number of drinking days. It did reduce drinks per drinking day and weekly craving, and predicted greater reductions in heavy drinking over time. In a subsample who smoked, it also predicted a relative reduction in cigarettes per day. The authors' own conclusion calls this initial prospective evidence justifying larger trials.
May 2026, 108 participants, 26 weeks
A single-center trial randomized 108 treatment-seeking participants with moderate to severe alcohol use disorder and comorbid obesity to semaglutide 2.4 mg or placebo for 26 weeks, on top of standard cognitive behavioral therapy for everyone[2]. Heavy drinking days fell 41.1 percentage points from baseline on semaglutide against 26.4 on placebo, an estimated treatment difference of 13.7 percentage points, p=0.0015. Eighty-one percent completed the full intervention.
Placebo plus cognitive behavioral therapy produced a 26.4-point drop in heavy drinking days on its own. The drug added 13.7 points on top of that. Both numbers belong in any sentence about this trial, and you can't quote the second without the first.
September 2026, 50 participants, oral semaglutide
The newest trial randomized 50 treatment-seeking adults with moderate to severe alcohol use disorder to oral semaglutide, 3 mg daily for four weeks then 7 mg for four weeks, or placebo[3]. It missed its primary outcome: laboratory cue-elicited craving at week six did not differ significantly, and neither did drinks per day. It did significantly reduce heavy drinking days, drinks per drinking day, naturalistic craving, alcohol-related consequences and cannabis use days, and more participants on drug dropped at least one World Health Organization risk drinking level.
PubMed, via the E-utilities API
(semaglutide[tiab] OR liraglutide[tiab] OR exenatide[tiab] OR dulaglutide[tiab] OR tirzepatide[tiab]) AND ("alcohol use disorder"[tiab] OR "alcohol consumption"[tiab] OR "heavy drinking"[tiab]) AND "randomized controlled trial"[pt] AND "humans"[mh]Returned 8. Positive control — naltrexone, identical filter — returned 201 through the identical filter in the same session.
Naltrexone is the control because it is the approved comparator a prescriber would actually weigh this against, and its trial record is twenty-five times the size. The eight include an exenatide trial from 2022 whose overall result was null, and a dulaglutide trial where alcohol was a secondary outcome during smoking cessation.
Naltrexone is worth pausing on for a second reason: it also turns up on these telehealth menus at a dose nothing approved supplies, for a different purpose entirely, and the evidence there went the other way. That story is in the low-dose naltrexone article.
What none of these trials is about
Every one of them enrolled people with diagnosed alcohol use disorder, two of them treatment-seeking, one of them alongside a course of therapy. None enrolled people taking a GLP-1 to lose weight and asking what a glass of wine does. The commonest way this research gets misread on a sales page is as a bonus effect for everyone on the drug, and the trials don't support that reading in either direction.
The approved labeling has less to say about drinking than most people expect. We read the full semaglutide prescribing information — 212,729 characters of it, in the version published June 30, 2026[4] — and counted every occurrence of the word.
The Wegovy prescribing information, full document text read the same day, version published June 30, 2026
full-text search of the label for "alcohol", each hit classified by hand
Returned 0. Positive control — hypoglycemia, same document, same read — returned 46 through the identical filter in the same session.
Eleven occurrences of the word in total: eight are instructions to clean the injection site with an alcohol swab, and three are the historical name of a liver disease in the phrase nonalcoholic steatohepatitis. None concerns drinking. The control shows the same search finds warning language in the same document when it is there, 46 times over. An absence on a label isn't a safety clearance; it means the question wasn't addressed there.
Whether a drug reduces how much someone drinks, and whether drinking while taking it is safe, are separate questions with separate evidence. The trials above answer the first one, partially, in a population with a diagnosis. The second belongs to whoever is prescribing, who can see your full medication list and history. Nobody here holds a medical license.
Why it matters on a telehealth intake
An intake form built around weight history doesn't necessarily reconcile a drinking history, and the questions it asks are one of the things we look at when we read a company. How these prescriptions get written, and by whom, is in the prescribing article; the frequency of everything else these drugs do is in the side-effect tables; and each company's own intake is described in its review.
Questions people actually ask
Does semaglutide reduce drinking?
In three randomized trials of people with alcohol use disorder, it reduced heavy drinking days and drinks per drinking day. It didn't reduce average drinks per calendar day in the 2025 trial or drinks per day in the 2026 oral trial, and two of the three missed or partly missed their own primary outcome. The largest enrolled 108 people.
Can I drink alcohol while taking a GLP-1?
That's a question for the prescriber who knows your medication list and history, and we aren't a clinical service. What we can report is that the approved semaglutide prescribing information does not address drinking: of eleven occurrences of the word alcohol in the full label, eight are injection-site swab instructions and three are part of a liver disease name.
Are GLP-1s approved to treat alcohol use disorder?
No. No GLP-1 receptor agonist carries an approved indication for alcohol use disorder. The published trials are phase 2 work in populations of 48, 50 and 108 people, and their own authors describe the results as justifying larger trials rather than as establishing a treatment.
How does this compare with naltrexone for drinking?
By volume of evidence, it doesn't compare yet. Eight randomized trials pair a GLP-1 with an alcohol outcome against 201 for naltrexone through the identical PubMed filter, and naltrexone has an approved indication for alcohol dependence while no GLP-1 does.
Sources
Every source here was fetched and read for this article, with the identifier taken off the record that came back and the claim it supports written down beside it. All of it was read in September 2026, the same session the rest of this page draws on.
- 1.Once-Weekly Semaglutide in Adults With Alcohol Use Disorder: A Randomized Clinical Trial. JAMA Psychiatry, 2025. PMID 39937469 · doi:10.1001/jamapsychiatry.2024.4789A phase 2 double-blind parallel-arm trial of 48 non-treatment-seeking adults with alcohol use disorder over nine weeks of semaglutide escalated to 1.0 mg; reductions in grams of alcohol consumed (-0.48) and peak breath alcohol concentration (-0.46) in a laboratory self-administration task; no effect on average drinks per calendar day or number of drinking days; significant reductions in drinks per drinking day and weekly craving; and the authors' conclusion that the findings justify larger trials.
- 2.Once-weekly semaglutide versus placebo in patients with alcohol use disorder and comorbid obesity: a randomised, double-blind, placebo-controlled trial. The Lancet, 2026. PMID 42070571 · doi:10.1016/S0140-6736(26)00305-3A 26-week single-center randomized trial of 108 treatment-seeking participants with moderate to severe alcohol use disorder and comorbid obesity receiving semaglutide 2.4 mg or placebo on top of standard cognitive behavioral therapy; heavy drinking days fell 41.1 percentage points against 26.4 on placebo, an estimated treatment difference of 13.7 percentage points, p=0.0015; 81% completed the full intervention.
- 3.Oral Semaglutide for Alcohol Use Disorder: A Randomized Clinical Trial. The American Journal of Psychiatry, 2026. PMID 42522065 · doi:10.1176/appi.ajp.20260003A phase 2 double-blind trial of 50 treatment-seeking adults randomized to oral semaglutide 3 mg daily for four weeks then 7 mg for four weeks, or placebo; no significant effect on the primary outcome of laboratory cue-elicited craving at week six or on drinks per day; significant reductions in heavy drinking days, drinks per drinking day, naturalistic craving, alcohol-related consequences and cannabis use days; and more participants reducing their World Health Organization risk drinking level by at least one level.
- 4.WEGOVY (semaglutide) injection, solution; WEGOVY (semaglutide) tablet — prescribing information. Novo Nordisk, via the DailyMed Structured Product Labeling service, 2026. Source · Document dated June 2026That the full prescribing information, 212,729 characters as read, contains eleven occurrences of the word alcohol — eight instructions to clean the injection site with an alcohol swab and three within the phrase nonalcoholic steatohepatitis — and none concerning alcohol consumption, while the same document uses the word hypoglycemia 46 times.
Key figures