This is one label's record. The class-wide picture — how Zepbound's figures sit next to Wegovy's, and why the discontinuation rates tell a different story from the nausea rates — is on our GLP-1 side effects page. This page reads the ZEPBOUND prescribing information (Eli Lilly, NDA 217806, SPL effective 28 August 2026) end to end, as published through openFDA and read on 23 September 2026, and reports what it counts. The same molecule's diabetes label, with different trial populations, is read on our Mounjaro side effects page.
This page reports a document; it does not advise. Where the Zepbound label instructs patients — on contraception, on planned procedures, on never sharing a pen — the instruction is quoted as the label's, and applying it is a matter for whoever prescribed the drug.
The counted reactions, with placebo subtracted
The label's Table 1 pools two placebo-controlled weight trials: Study 1 (SURMOUNT-1, adults without diabetes) and Study 2 (SURMOUNT-2, adults with type 2 diabetes). In total 2,519 people received Zepbound for up to 72 weeks. The last column is our arithmetic, not the label's: the 15 mg rate minus the placebo rate, which is the share of people for whom the drug, rather than ordinary life, plausibly explains the report.
| Reaction | Placebo | 5 mg | 10 mg | 15 mg | Excess at 15 mg |
|---|---|---|---|---|---|
| Nausea | 8% | 25% | 29% | 28% | +20 |
| Diarrhoea | 8% | 19% | 21% | 23% | +15 |
| Vomiting | 2% | 8% | 11% | 13% | +11 |
| Constipation | 5% | 17% | 14% | 11% | +6 |
| Abdominal pain | 5% | 9% | 9% | 10% | +5 |
| Indigestion (dyspepsia) | 4% | 9% | 9% | 10% | +6 |
| Injection-site reactions | 2% | 6% | 8% | 8% | +6 |
| Fatigue | 3% | 5% | 6% | 7% | +4 |
| Hypersensitivity reactions | 3% | 5% | 5% | 5% | +2 |
| Burping (eructation) | 1% | 4% | 5% | 5% | +4 |
| Hair loss | 1% | 5% | 4% | 5% | +4 |
| Reflux | 2% | 4% | 4% | 5% | +3 |
| Flatulence | 2% | 3% | 3% | 4% | +2 |
| Abdominal distension | 2% | 3% | 3% | 4% | +2 |
| Dizziness | 2% | 4% | 5% | 4% | +2 |
| Low blood pressure | 0% | 1% | 1% | 2% | +2 |
Placebo is not zero, and for several rows it is most of the number. Eight in every hundred people injecting nothing reported nausea. A reader trying to decide whether a symptom is the drug is working against that background. Tiredness is a good example: 3% on placebo against 7% at 15 mg, and Zepbound's fatigue row pools four terms where some labels count one — compared label by label on does GLP-1 make you tired.
The number that does not move, and the one that does
The label adds a figure the table cannot show: the share reporting any gastrointestinal reaction was 56% on 5 mg, 56% on 10 mg and 56% on 15 mg, against 30% on placebo. Tripling the dose did not change how many people had a stomach complaint at all.
What did change was how many stopped because of one: 1.9% at 5 mg, 3.3% at 10 mg, 4.3% at 15 mg, against 0.5% on placebo. The count of people affected is flat; the count of people affected badly enough to quit more than doubles. On this label the dose effect is carried by severity, not by frequency — which is why a single "nausea percentage" is the least informative number on the page.
Overall, 4.8%, 6.3% and 6.7% of patients on the three doses permanently stopped for any adverse reaction, against 3.4% on placebo. The label says most who stopped did so "during the first few months of treatment due to gastrointestinal adverse reactions", and that most nausea, vomiting and diarrhoea "occurred during dose escalation and decreased over time". It gives no duration for an individual episode. How that stopping rate sits beside the other weight labels, and what the direct comparisons between drugs found, is on which GLP-1 has the fewest side effects. How Zepbound's diarrhoea rate compares with eight other labels, placebo for placebo, is on diarrhoea across the GLP-1 labels.
One more population, a less flattering one. Study 3 (SURMOUNT-3) started people on an intensive lifestyle programme before they received anything. After that lead-in, 10% of Zepbound patients and 2% on placebo stopped because of adverse reactions — a higher rate than in the pooled trials. The label reports it and does not comment. People who had already lost weight on diet alone before starting were, in that trial, more likely to stop than the pooled figures suggest.
The one reaction that falls with dose, constipation, is discussed on our GLP-1 side effects page; the label offers no explanation for it and neither will we.
The antibody effect
Section 6 reports something the headline table flattens. Among Zepbound patients in the pooled trials:
| Developed anti-tirzepatide antibodies | Did not | |
|---|---|---|
| Injection-site reactions | 11.3% | 1% |
| Hypersensitivity reactions | 6.2% | 3% |
Injection-site reactions were about eleven times as common in people whose immune system made antibodies to the drug — the largest such gap in any GLP-1 label, set beside the others on GLP-1 injection-site reactions. Most hypersensitivity reactions in the trials were skin reactions such as rash and itching; immediate reactions within a day of an injection occurred in 2.1% on Zepbound against 0.4% on placebo. The label does not say what share of patients develop antibodies in this section, and does not suggest any action based on antibody status — nobody is routinely tested for them. Hypersensitivity away from the injection site, mostly rash and itching in the trials, is tabled against the other labels on Zepbound rash and itching.
Blood sugar, blood pressure, heart rate
- Low blood sugar. In the diabetes trial (Study 2), a reading below 54 mg/dL was reported in 4.2% on Zepbound against 1.3% on placebo. In Study 1, which excluded diabetes and did not systematically capture it, 0.3% against none. The label's warning is chiefly about use alongside insulin or a sulfonylurea.
- Low blood pressure. 1.6% against 0.1%, and 2.2% in people already taking blood-pressure medicines against 1.2% in those who were not. Some episodes were linked to stomach upset and dehydration. The other side of the record — systolic pressure falling 4 to 7 mmHg more than placebo in the label's trials — is compared across brands on GLP-1 drugs and blood pressure.
- Heart rate. A mean increase of 1 to 3 beats per minute, against none on placebo.
- Kidney injury. 0.5% against 0.2%. The postmarketing section adds acute kidney failure or worsening chronic kidney failure, "sometimes requiring hemodialysis".
The contraception warning
Of everything in this label, this is the sentence most likely to matter to a reader and least likely to have been mentioned to them. Section 7 advises patients using oral hormonal contraceptives to switch to a non-oral method, or add a barrier method, for 4 weeks after starting and for 4 weeks after each dose increase. The reason given is delayed stomach emptying, which can reduce how much of an oral medicine is absorbed. The label states that hormonal contraceptives not taken by mouth should not be affected.
Two things follow from the label's own wording. The window reopens at every dose step, not only at the start. And because the Zepbound label's section 8 carries its own pregnancy guidance, this is not a footnote; our pregnancy page sets out what the labels say about conception and discontinuation. Whether any of it applies to a particular person's contraception is a question for their prescriber or pharmacist.
The sleep-apnoea population
Zepbound is also approved for moderate to severe obstructive sleep apnoea in adults with obesity, based on Studies 5 and 6 (SURMOUNT-OSA; 467 adults). The label says the side effects in those trials were similar to the weight trials, with one figure that is not: adjudication-confirmed acute pancreatitis at 0.84 patients per 100 years of exposure on Zepbound and 0 on placebo. In the weight trials the same figure was 0.14 against 0.15. The OSA trials are small, and a rate built on a few cases moves a long way; the label reports it without interpretation, and so do we. The full pancreatitis record, across the class, is on our pancreatitis page.
The label folds the two OSA trials into one sentence ("similar" to the weight trials). The registry record of the master protocol, NCT05412004 (results first posted 30 April 2025, read 14 October 2026), keeps them apart. Study 5 enrolled people not using a PAP machine, Study 6 people who were; both gave the highest tolerated dose, 10 or 15 mg, for 52 weeks. Percentages are our division of the posted counts:
| Reported in the registry | Study 5, no PAP: Zepbound (114) | Placebo (120) | Study 6, on PAP: Zepbound (119) | Placebo (114) |
|---|---|---|---|---|
| Nausea | 25.4% | 10.0% | 21.8% | 5.3% |
| Diarrhoea | 26.3% | 12.5% | 21.8% | 8.8% |
| Vomiting | 17.5% | 4.2% | 9.2% | 0.9% |
| Constipation | 15.8% | 2.5% | 15.1% | 4.4% |
| Any serious adverse event | 7.9% | 5.8% | 5.9% | 10.5% |
Two things in that table are not in the label. Vomiting in the no-PAP trial (17.5%) is higher than any dose in the weight trials' Table 1 (8–13%), though the populations, the 52-week length and the counting rules all differ, so the figures are set side by side, not compared. And serious adverse events ran the other way in the PAP trial — more on placebo than on Zepbound — which is what a small trial with a few dozen serious events looks like, not evidence of protection. The registry's participant flow counts people who left the study citing an adverse event: 0 and 1 on Zepbound, 2 and 5 on placebo; that is leaving the study, a narrower thing than stopping the drug, and the label's discontinuation figures for these trials are not printed.
The warnings, and where each is covered
The label's Warnings and Precautions list nine items plus a boxed warning. We do not restate the ones that already have their own page.
- Thyroid C-cell tumours (boxed warning, from rat studies) — thyroid cancer page.
- Severe gastrointestinal reactions, and the label's statement that Zepbound is not recommended in severe gastroparesis.
- Acute kidney injury from dehydration — above.
- Acute gallbladder disease — cholelithiasis 1.1% against 1%, cholecystitis 0.7% against 0.2%; gallbladder page.
- Acute pancreatitis — pancreatitis page.
- Hypersensitivity, including anaphylaxis and angioedema after approval.
- Hypoglycaemia — above.
- Diabetic retinopathy complications in people with type 2 diabetes.
- Pulmonary aspiration during anaesthesia or deep sedation — the label tells patients to inform their healthcare providers of planned procedures; our surgery page covers the guidance.
- Never share a KwikPen, even with a new needle.
Hair loss (4–5% against 1%, and 7.1% of women against 0.5% of men) is on our hair loss page. How these rates change as the dose climbs, across the class, is on our dose and side effects page.
What the Zepbound label cannot tell you
- Anything past 72 weeks. SURMOUNT-1 and -2 ran up to 72 weeks and the sleep-apnoea trials 52; the label holds no Zepbound side-effect table for longer use, and a reaction that appears only with years of treatment would not be in it.
- Anyone the trials left out. The weight trials enrolled adults with obesity or overweight plus a weight-related condition, under study-visit supervision and a reduced-calorie diet; the label's figures say nothing about people outside those criteria, or about doses taken outside the 5, 10 and 15 mg arms.
- A comparison with Mounjaro, Wegovy or any other label. The label itself says its rates cannot be set against another programme's; even the same molecule prints different numbers in its diabetes label, as the Mounjaro comparison shows.
- How often the postmarketing harms happen. Pancreatitis sometimes resulting in death, bowel obstruction, severe constipation, anaphylaxis and kidney failure are listed from reports in "a population of uncertain size" — named, not counted.
A symptom on Zepbound, or a worry about one, belongs with the prescriber, who has the patient's history this label does not. The efficacy side of the same trials is in our tirzepatide review.
Sources
- ZEPBOUND (tirzepatide) injection, prescribing information, Eli Lilly and Company, NDA 217806, SPL effective 2026-08-28 — read 2026-09-23 via openFDA: boxed warning, sections 1, 5, 6.1, 6.2 and 7.
- SURMOUNT-OSA master protocol NCT05412004, ClinicalTrials.gov API v2 results (adverse events by trial and participant flow), results first posted 2025-04-30 — read 2026-10-14. Label re-read via openFDA the same day: SPL effective 2026-08-28, version 40, unchanged.
- SURMOUNT-1 NCT04184622 and SURMOUNT-2 NCT04657003, as pooled in the label's Table 1.