Short answer: the evidence runs 176 weeks, and almost nothing new appeared after the first 20
The longest randomised, placebo-controlled evidence on tirzepatide in obesity runs 176 weeks, about three years and four months. That is the SURMOUNT-1 continuation, published in the New England Journal of Medicine in 2025, following 1,032 participants with obesity and prediabetes on drug or placebo for 176 weeks and then 17 weeks off it. Its safety conclusion was specific: aside from COVID-19, the most common adverse events were gastrointestinal, mostly mild to moderate, and they "occurred primarily during the dose-escalation period in the first 20 weeks of the trial". No new safety signals were identified. So the honest answer to what the long-term side effects are is that, out to three years, they are mostly the same side effects as the short-term ones, arriving early rather than late. What genuinely remains open is listed further down, and it is shorter and stranger than most pages suggest.
"Long-term side effects" is a question about time, so an answer that does not tell you how much time the evidence covers has not answered it. Most pages on this query list the same risks that appear on every Zepbound page, add a sentence about how the drug is new, and stop. The better competitors cite a 36 week maintenance trial. The data now runs roughly five times longer than that, so this page is organised by evidence horizon first and risk second.
How far the evidence actually runs
| Study | Randomised duration | Participants | What it adds on duration |
|---|---|---|---|
| SURMOUNT-1 primary analysis | 72 weeks | 2,539 | The adverse reaction table on the label comes from this and its companion |
| Label Study 4, randomised withdrawal | Up to 36 weeks, then up to 88 weeks in a subset | 783, of whom 335 reached 88 weeks | The label states reactions here were similar to the pivotal pool |
| SURMOUNT-4 | 36 weeks open label, then 52 weeks randomised withdrawal | 670 randomised | What happens on stopping, which is a different question from accumulating harm |
| SURMOUNT-1 continuation | 176 weeks, plus 17 weeks off treatment | 1,032 with obesity and prediabetes | The longest randomised comparison. No new safety signals |
Three years of randomised data is a great deal more than most newly approved medicines have, and considerably less than the lifetime many people expect to take this for. Both of those statements are true at once, and any page that gives you only one of them is selling something.
Three categories, because they behave differently over time
Category one: the ones that front-load
The gastrointestinal reactions are the bulk of the experience, and the label and the three year follow up agree on their shape. From the label: the majority of nausea, vomiting and diarrhea occurred during dose escalation and decreased over time. From the 176 week report: adverse events occurred primarily in the first 20 weeks. These are not long-term side effects in any meaningful sense. They are the price of the first few months.
Category two: the ones tied to losing weight, which continue while you do
Several documented reactions are consequences of the weight loss rather than of time on the drug, and the label says so directly in two places.
| Event | Zepbound | Placebo | What the label says about why |
|---|---|---|---|
| Cholelithiasis (gallstones) | 1.1% | 1.0% | Acute gallbladder events were associated with weight reduction |
| Cholecystitis | 0.7% | 0.2% | Same |
| Cholecystectomy | 0.2% | none | Same |
| Hair loss | 5% (women 7.1%, men 0.5%) | 1% | Hair loss reactions were associated with weight reduction |
| Acute kidney injury | 0.5% | 0.2% | Mostly in patients dehydrated by gut reactions |
The practical reading is that these follow the pace of your weight change rather than a counter of weeks on treatment, which means they are most relevant during active loss and less so during maintenance.
Category three: the ones with no long-term human data at all
This is the honest core of the question, and it is small.
- Thyroid C-cell tumours. The boxed warning rests on a two year study in rats, in which tirzepatide caused a dose-dependent and duration-dependent increase in thyroid C-cell tumours. The label states that it is unknown whether Zepbound causes these tumours in humans, because the human relevance of the rodent finding has not been determined. It also says routine calcitonin monitoring or thyroid ultrasound is of uncertain value, and may cause unnecessary procedures. Zepbound is contraindicated for anyone with a personal or family history of medullary thyroid carcinoma or with MEN 2. This is the one risk where "long-term" genuinely means unresolved.
- Pancreatic enzyme drift. Treatment raised mean pancreatic amylase by 20% to 25% and lipase by 28% to 35% from baseline, against 2.1% and 5.8% on placebo. The label's own comment on this is that the clinical significance is unknown in the absence of other signs of pancreatitis. Adjudicated acute pancreatitis itself was 0.2% on drug and 0.2% on placebo in the pivotal pool.
- Physical function after years of lean mass change. A 2026 systematic review in Annals of Internal Medicine covering 35 trials found that none of them reported an objective physical function outcome. The body composition data exists; whether it translates into measurable loss of strength or capacity has not been published.
Two claims we went looking for and could not source
Bone density and osteoporosis appear in several articles on this query as long-term Zepbound risks. We could not find them in the prescribing information or in a trial report, so they are not on this page. Vision is more subtle: the label does warn about temporary worsening of diabetic retinopathy with rapid glucose improvement, and asks that patients with a history of diabetic retinopathy be monitored. That is a specific warning for people who already have that condition, not a general long-term vision risk, and it is often quoted as though it were the latter.
What three years of data did show about stopping
The 176 week trial included a 17 week off-treatment period, and that is where the most load-bearing long-term finding sits. During treatment, 1.3% of participants on tirzepatide received a diagnosis of type 2 diabetes against 13.3% on placebo. After the 17 weeks off, the figures were 2.4% and 13.7%. The separate SURMOUNT-4 trial tested the same principle on weight. After a 36 week lead-in, 670 participants were randomised to continue tirzepatide or switch to placebo for 52 weeks; weight changed by -5.5% on continued treatment and +14.0% on placebo over that period. Together those say that this is treatment for a chronic condition rather than a course you complete, which is the real long-term consideration for most people and has nothing to do with toxicity.
What a long-term record is actually for
Three years of data in a trial does not tell you about your third year. These are the things that only your own record can answer, and they are the ones a prescriber asks about at a review:
- Whether anything genuinely new appeared after month six, as distinct from something that never went away.
- Whether a symptom still tracks to the dose, which is the test that distinguishes a drug effect from life.
- What happened around each dose change, including any held or reduced dose.
- How the rate of weight change moved, since the gallbladder and hair rows above follow that rather than the calendar.
A record that is still useful in year three
Jabby keeps dose history, injection sites and symptoms logged against the dose they followed, so a long-term picture exists when someone asks for one.
Download Jabby, free on the App StoreWhen something new is not a long-term side effect to monitor
Contact your prescriber or seek urgent care for persistent severe abdominal pain, which may radiate to the back, with or without vomiting; for a lump in the neck, trouble swallowing, shortness of breath or persistent hoarseness, which the label names as symptoms to report because of the thyroid warning; for signs of a serious allergic reaction; or for symptoms of gallbladder disease. Tell any clinician arranging surgery or a procedure under general anaesthesia or deep sedation that you take this medicine.
Frequently asked questions
What are the long-term side effects of Zepbound?
Out to 176 weeks in a randomised trial, no new safety signals appeared, and adverse events were predominantly gastrointestinal and concentrated in the first 20 weeks. The risks that persist are the ones tied to weight loss itself, such as gallbladder events and hair loss, and the ones with unresolved human data, principally the rodent thyroid C-cell finding behind the boxed warning.
How long has Zepbound been studied for?
The longest randomised, placebo-controlled obesity data is 176 weeks of treatment plus 17 weeks off it, in 1,032 participants with obesity and prediabetes. The label's own longest cited exposure is up to 88 weeks, in 335 patients in its Study 4.
Do side effects get worse the longer you take it?
That is not the pattern in the published data. The label records that the most common reactions decreased over time after dose escalation, and the three year report placed most events in the first 20 weeks.
Is the thyroid cancer warning based on human data?
No. It comes from a two year study in rats. The label states explicitly that it is unknown whether Zepbound causes thyroid C-cell tumours in humans, and that the human relevance of the rodent finding has not been determined. It is still a contraindication for anyone with a personal or family history of medullary thyroid carcinoma or with MEN 2.
Does Zepbound cause bone loss?
We could not find bone density or osteoporosis in the prescribing information or in a trial report, so we are not repeating that claim. If you have specific bone health concerns, raise them with your clinician rather than acting on an unsourced list.
Is it safe to stay on it indefinitely?
That is a question for your prescriber, and the label describes long-term weight maintenance as part of the approved use. What the data shows is that the benefits measured, on weight and on progression to type 2 diabetes, narrowed after stopping, which is why continuation is a clinical conversation rather than an automatic choice in either direction.
Related reading
- How do you reduce the side effects of Zepbound?, which matters most in the first 20 weeks this page describes
- Do Zepbound side effects start right away?, the other end of the same timeline
- Zepbound side effects and muscle pain, including the lean mass data behind the physical function gap
- Does Zepbound cause joint pain?, another symptom absent from the label
- Zepbound reviews, and the four things no review can tell you
- Zepbound injection site reactions, the local reactions by dose
- Tracking tirzepatide injections, for keeping a record that survives three years
Sources
- Jastreboff AM, le Roux CW, Stefanski A, et al. Tirzepatide for obesity treatment and diabetes prevention. N Engl J Med 2025;392(10):958-971. PMID 39536238. The 176 week randomised data, the first 20 weeks finding, the diabetes diagnosis rates and the 17 week off-treatment period. Registration NCT04184622.
- ZEPBOUND (tirzepatide) US prescribing information, Eli Lilly and Company. Boxed warning and section 5.1 for the rodent thyroid C-cell study, 5.4 for gallbladder, 5.5 for pancreatitis, 5.8 for diabetic retinopathy, 6.1 for every percentage above including the amylase and lipase changes and the Study 4 exposure figures. PDF version.
- Aronne LJ, et al. Continued treatment with tirzepatide for maintenance of weight reduction in adults with obesity: the SURMOUNT-4 randomized clinical trial. JAMA 2024;331(1):38-48. PMID 38078870. Registration NCT04660643.
- Jastreboff AM, et al. Tirzepatide once weekly for the treatment of obesity. N Engl J Med 2022;387:205-216. PMID 35658024. The 72 week primary analysis.
- Batsis JA, Gavras A, Gross DC, et al. Effect of incretin-based and nonpharmacologic weight loss on body composition: a systematic review. Ann Intern Med 2026;179(7):996-1013. PMID 41996180. Source of the absence of physical function outcomes across 35 trials.
- ZEPBOUND label on DailyMed, US National Library of Medicine.
This page summarises published prescribing information and peer-reviewed trial reports for general education. It is not medical advice, it contains no dosing guidance, and absence of a risk from a label or a trial is not a guarantee of safety. Decisions about continuing, pausing or stopping long-term treatment belong with your prescriber. Never start, stop or change a prescribed medicine on the basis of a web page.
Jabby is free on the App Store and keeps dose history and symptom logs in one record you can show your prescriber.