The question usually arrives in a simple form: is it dangerous to use cannabis while taking semaglutide or tirzepatide? The direct answer is that no pharmacological interaction between the two has been described in the prescribing information or measured in a trial. The useful answer is longer, because both act on the same organs. Each slows the stomach, each pushes appetite, in opposite directions, and each is linked to a vomiting problem. The parallel question about drinking is covered in the alcohol article, and the overlap with cannabis runs through different physiology.
The labels are silent on cannabis and specific about the stomach
The current prescribing information for Wegovy and for Zepbound, as published on DailyMed, does not contain the words cannabis, marijuana, cannabinoid or dronabinol anywhere: not in the warnings, not in the drug interaction section, not in the patient counseling text.[1][2] No interaction entry exists to report.
That silence is a statement about what was studied, not a clearance. What both labels do record is the mechanism the overlap runs through. Each states that the drug delays gastric emptying and may affect the absorption of oral medications taken with it. The tirzepatide label goes further, advising people on oral hormonal contraceptives to switch methods or add a barrier method for four weeks after starting and after each dose increase.[2] Edible cannabis is an oral product, and no study has measured whether a slowed stomach changes how it is absorbed.
Two substances that both hold food in the stomach
The cannabis half of that overlap has been measured directly. In a double-blind, randomized crossover study of 13 healthy volunteers, oral delta-9-tetrahydrocannabinol (THC) at 10 mg/m² slowed the emptying of a radiolabeled solid meal in every subject. Two hours after eating, 73.9% of the meal remained in the stomach after THC, against 45.6% after placebo (P < 0.001).[3]
That is a small, old study, and the dose was the one used for chemotherapy nausea rather than a typical recreational amount. It is nonetheless the same direction of effect the GLP-1 labels describe, and no study has measured the two together. Whether the effects add, and by how much, is unknown. People with existing slow emptying have more to weigh, and that condition is covered in the gastroparesis article.
The munchies are a measured effect, and so is the reverse
The appetite effect is the most intuitive conflict and also one of the better documented. In a residential laboratory study from 1986, nine men smoked active or placebo cigarettes under continuous observation. A single cigarette before a work period did not change food intake. Two or three cigarettes during social time raised average daily calorie intake, and the extra calories came from snacks between meals rather than larger meals.[4]
A 2026 randomized crossover trial added a result that complicates the assumption that THC alone drives eating. Fifteen healthy adults took 298 mg of cannabidiol (CBD) or placebo. At an all-you-can-eat lunch three hours later, they ate 193 kcal more after CBD (95% CI, 80 to 306), even though ghrelin was lower and subjective appetite ratings did not differ.[5] The non-intoxicating compound often sold as a wellness product raised intake without anyone feeling hungrier.
Neither study involved a GLP-1 drug, so neither shows that cannabis blunts weight loss on one. What they show is that cannabis can raise intake through grazing and through a route people do not notice, and that is the pattern GLP-1 drugs are bought to quiet. The appetite side of that purchase is examined in the food noise article.
Vomiting that does not stop: two syndromes that look alike
Nausea and vomiting are the signature side effects of the drug class, and their dose pattern is set out in the vomiting article. Heavy cannabis use carries its own vomiting disorder. Cannabinoid hyperemesis syndrome (CHS) presents as recurrent episodes of severe vomiting, and a 2025 review notes that most patients have used cannabis daily or near-daily for more than two years. Compulsive hot showers or baths are reported by most patients but are not specific to the condition. Sustained abstinence is the cornerstone of treatment.[6]
The syndrome is becoming more common in emergency departments. In Nevada, CHS visits rose from 1.07 to 2.22 per 100,000 after commercial recreational sales began in 2017.[7] The practical problem for a GLP-1 user is attribution. Vomiting after a dose increase looks like the drug; vomiting after years of daily use looks like CHS; in a daily user who just raised a dose, both explanations fit. The same review describes patients’ skepticism that cannabis could cause their symptoms as one of the main obstacles to treatment.[6] No study has measured how often the two overlap. Typical GLP-1 nausea follows a predictable timeline, described in the nausea timeline, and vomiting that falls outside it deserves a full history, including cannabis use.
Sedation, procedures and a stomach that has not emptied
Both labels carry the same warning: rare postmarketing reports of pulmonary aspiration during general anesthesia or deep sedation in patients on GLP-1 drugs who had food left in the stomach despite fasting as instructed. The Wegovy label states that available data are insufficient to recommend whether changing fasting rules or pausing the drug reduces that risk, and tells patients to inform providers before any planned procedure.[1]
Cannabis enters the same room separately. A 2025 review reports that chronic cannabis use is associated with higher propofol doses during sedation and general anesthesia, although no evidence yet shows causality. It summarizes current guidance as asking every patient about cannabis use before surgery and postponing elective procedures when a patient is acutely intoxicated.[8] The aspiration evidence for GLP-1 drugs themselves, which is weaker than the warning suggests, is in the aspiration article, and the pre-procedure guidance is in the surgery article. The practical point is to disclose both, since each changes what an anesthesiologist plans for.
The cohort finding that made headlines
The research that made this question popular runs the other way: whether GLP-1 drugs reduce problem cannabis use. A 2024 retrospective cohort drew on health records from about 105 million patients and compared semaglutide with non-GLP-1 weight or diabetes medicines over 12 months. Among 85,223 patients with obesity, semaglutide was associated with a lower risk of a first cannabis use disorder diagnosis, at HR 0.56 (95% CI, 0.42 to 0.75), and of a repeat diagnosis in people with a prior one, at HR 0.62 (0.46 to 0.84).[9]
The replication in 596,045 patients with type 2 diabetes produced a stronger figure for new diagnoses, HR 0.40 (0.29 to 0.56). For repeat diagnoses, though, it reached only HR 0.66 (0.42 to 1.03), an interval that includes no effect.[9] The result most relevant to treatment, preventing relapse, is the one that did not clear significance in the larger population. The authors call the findings preliminary and say randomized trials are needed.
A 2026 target trial emulation in US veterans with type 2 diabetes compared 124,001 people starting a GLP-1 drug with 400,816 starting an SGLT-2 inhibitor. New cannabis use disorder was less frequent with the GLP-1, at HR 0.86 (0.81 to 0.90). In absolute terms, that is 2.25 fewer diagnoses per 1,000 people over three years.[10] The direction matches the 2024 study; the size is far smaller once the comparator is another modern diabetes drug.
The numbers that complicate the reading
Genetics does not support a cannabis-specific effect. A 2025 multi-ancestry Mendelian randomization study found genetic proxies for GLP-1 and GIP receptor activity associated with less heavy drinking, while estimates for tobacco, cannabis and opioid use disorders were consistently null.[11]
The sharpest reversal comes from psychiatric outcomes. In a 2025 records study of 6.4 million adults with type 2 diabetes, GLP-1 use was associated with fewer suicide attempts overall, at aHR 0.63 (0.47 to 0.85). Cannabis use disorder itself carried an aHR of 5.50. Among people with cannabis use disorder, GLP-1 users and non-users did not differ: aHR 1.00 (0.37 to 2.69).[12] The apparent protection disappeared in the group with the most risk. The wider mood evidence is in the mental health article.
A 2026 cohort asked the opposite question: whether cannabis use changes outcomes for people starting a GLP-1. Among 4,117 matched pairs with type 2 diabetes, people with cannabis-use diagnoses had higher all-cause mortality, at HR 1.64 (1.38 to 1.96), and more major cardiovascular events (HR 1.42) and kidney events (HR 1.55). Risk was higher with cannabis use disorder (HR 2.20) than with non-disordered use (HR 1.73).[13] That study cannot separate the drug from the lives of the people who use it heavily, but its conclusion is direct: prescribers should ask about cannabis.
All of these studies are observational, and all rely on diagnosis codes. People who are never coded for cannabis use, which describes most occasional users, are absent from them. The broader addiction picture, including the trials that missed their endpoints, is in the alcohol use disorder article, and the tobacco evidence is in the smoking article.
What a compounded prescription adds
Every study above used branded products. Compounded semaglutide and tirzepatide are not FDA-approved, and the FDA does not review them for safety, effectiveness or quality before they are dispensed. The gastric-emptying and aspiration warnings come from labels written for the branded drugs. A compounded vial carries no equivalent prescribing information. The differences are covered in the compounded versus brand article.
What this leaves
No interaction is listed, and none has been measured. The concerns that have evidence are physiological and practical: two substances that slow the stomach, an appetite effect that works against the treatment, a vomiting syndrome that can be mistaken for a side effect, and a procedure risk that both affect. The idea that a GLP-1 treats cannabis dependence rests on observational associations that weaken against an active comparator and disappear in the highest-risk group. The useful step is to tell the prescriber about cannabis use, and an intake that never asks is one of the patterns in the telehealth article.