There are two ways to read a drug label's side-effect section. Most coverage reads the incidence table, quotes the largest number in it, and stops. The more useful number is four paragraphs further down, and on these two drugs it says something the incidence tables do not.
Every figure below was read on 2026-09-10 from the FDA labels published through openFDA: the Wegovy label (the manufacturer's current label, effective 2026-06-18, re-read 2026-09-23; it prints the same 2.4 mg figures as the older copy first read, and is covered in full on our Wegovy side effects page), the Zepbound label effective 2026-08-28, and for the postmarketing comparison the Ozempic label effective 2026-01-30, the Mounjaro label effective 2026-08-27 (Eli Lilly's copy, re-checked 2026-09-24) and the Saxenda label effective 2026-02-25. This page reports what those documents record. It does not say who should take these medicines or what anyone should do about a symptom; that belongs with a prescriber, as our editorial standards set out.
Counted reactions are one half of a label's safety documentation. The other half — the short list of people for whom each label rules the drug out entirely, and the much longer list of what it says to watch — is set out in who the labels contraindicate, where one product turns out to carry neither the class boxed warning nor the thyroid restriction. Two diabetes labels have record pages of their own: Mounjaro's counted reactions and Ozempic's, with its eye warning worked through. One rare effect sits outside the US labels altogether: the NAION record, where European and UK regulators have acted.
The tables everyone quotes
Both weight-management labels report the same kind of table: adverse reactions occurring in at least a set percentage of patients in the pooled placebo-controlled trials, with the placebo column beside them. The placebo column is the part that does the work, because it is the only thing that separates a side effect from an ordinary week.
Wegovy (semaglutide 2.4 mg), pooled placebo-controlled trials, N = 2,116
| Reaction | Placebo | Wegovy |
|---|---|---|
| Nausea | 16% | 44% |
| Diarrhoea | 16% | 30% |
| Vomiting | 6% | 24% |
| Constipation | 11% | 24% |
| Abdominal pain | 10% | 20% |
| Headache | 10% | 14% |
| Fatigue | 5% | 11% |
| Dyspepsia | 3% | 9% |
| Dizziness | 4% | 8% |
| Abdominal distension | 5% | 7% |
| Belching | under 1% | 7% |
| Hair loss | 1% | 3% |
Abdominal pain, dyspepsia and reflux are read together, with the cohort studies behind the reflux signal, on GLP-1 acid reflux and heartburn.
Zepbound (tirzepatide), pooled placebo-controlled trials for weight reduction
| Reaction | Placebo (N=958) | 5 mg (N=630) | 10 mg (N=948) | 15 mg (N=941) |
|---|---|---|---|---|
| Nausea | 8% | 25% | 29% | 28% |
| Diarrhoea | 8% | 19% | 21% | 23% |
| Constipation | 5% | 17% | 14% | 11% |
| Vomiting | 2% | 8% | 11% | 13% |
| Abdominal pain | 5% | 9% | 9% | 10% |
| Dyspepsia | 4% | 9% | 9% | 10% |
| Injection site reactions | 2% | 6% | 8% | 8% |
| Fatigue | 3% | 5% | 6% | 7% |
| Belching | 1% | 4% | 5% | 5% |
| Hair loss | 1% | 5% | 4% | 5% |
Two things in those tables are worth pausing on before the comparison that follows.
The first is that nausea on placebo is not zero. Sixteen percent of people who injected nothing at all in the Wegovy trials reported nausea, and 11% reported constipation. A reader working out whether a symptom is the drug is working against a background rate that is substantial and, in the Wegovy trials, twice the Zepbound trials' background rate — 16% against 8% for the same symptom in the same kind of study.
The second is that constipation is the one reaction in the Zepbound table that goes down as the dose goes up: 17% at 5 mg, 14% at 10 mg, 11% at 15 mg. Everything else rises or holds. The label offers no explanation and this page will not invent one.
The number the tables do not give you
Here is the comparison that changed how we read both documents.
| Wegovy 2.4 mg | Zepbound 15 mg | |
|---|---|---|
| Nausea reported | 44% | 28% |
| Placebo nausea | 16% | 8% |
| Permanently discontinued for adverse reactions | 6.8% | 6.7% |
| Placebo discontinuation | 3.2% | 3.4% |
The nausea figures are sixteen percentage points apart. The discontinuation figures are one tenth of a point apart, on placebo rates that differ by two tenths of a point.
That is not what a reader would predict from the headline numbers, and it is the single most useful thing in either label. Reporting a symptom in a trial diary and abandoning a treatment are different events, and the two labels agree far more closely on the second than on the first. Whatever is producing the gap in reported nausea — different trial populations, different diary instruments, different tolerance thresholds among people recruited into a semaglutide study rather than a tirzepatide one — it is not showing up as a difference in how many people quit. How long a single episode of nausea lasts, and at what point in treatment it peaks, is set out on our page on GLP-1 nausea. Trulicity's own table, which reports only its two original strengths, is on Trulicity: label and side effects.
The honest limitation, stated plainly: these are not head-to-head trials. Wegovy's pooled figures and Zepbound's pooled figures come from separate studies with separate participants, and no amount of table-alignment makes them a comparison. The placebo columns are the closest thing to a common yardstick, and they differ too — 16% against 8% for nausea. What survives that caveat is the internal comparison inside each label: in both documents, the proportion of people who stopped is a small fraction of the proportion who reported the symptom.
Wegovy's label goes one step further and names the reactions that actually ended treatment, which almost nothing published about this drug quotes: nausea 1.8% against 0.2% on placebo, vomiting 1.2% against 0%, diarrhoea 0.7% against 0.1%. Forty-four percent of patients reported nausea. Under two percent stopped because of it.
What happens early
Zepbound's label carries one sentence about timing: the majority of patients who discontinued because of adverse reactions did so during the first few months of treatment. A month-by-month view across a year, and how long each drug stays after the last dose, is on the GLP-1 side-effect duration page.
That sentence is usually read as reassurance — the hard part is at the beginning. It supports a second reading as well, and a careful reader should hold both. A trial population that loses its least tolerant participants in the first few months will report lower rates in later months partly because those people are no longer in it. The label does not distinguish between symptoms fading and intolerant patients leaving, and neither does any figure in it. This page is not going to claim the first when the document supports both.
The obvious follow-up is how much of the burden belongs to the dose rather than to the drug. Zepbound's label answers it with three dose columns, and Wegovy's current label, effective 18 June 2026, adds a 7.2 mg arm beside 2.4 mg. What they show when the placebo column is subtracted is that tripling an escalated dose moves excess nausea far less than the dose — and that the steepest gradient on Wegovy is a skin-sensation reaction, not a stomach one.
The one that depends on somebody else knowing
Across all five labels checked — Wegovy, Ozempic, Zepbound, Mounjaro and Saxenda — one postmarketing entry appears in the same form every time: pulmonary aspiration during general anaesthesia or deep sedation, reported in patients receiving GLP-1 receptor agonists undergoing elective surgeries or procedures, with an instruction to inform healthcare providers about the medicine before a procedure.
It is a rare event. The labels attach no frequency to it, which is standard for postmarketing reports, where the denominator is unknown. It is also the one entry on these lists where the labels explicitly refuse to say what should be done about it, and where anaesthesia societies in different countries have reached opposite conclusions from the same thin evidence. What makes it worth a section here is structural rather than clinical: it is the only item on any of these lists whose management depends on information travelling from a patient to an anaesthetist. Every other entry in the tables above is something a person experiences and can describe. This one is a fact about a prescription that has to be spoken aloud in a different room, weeks or months later.
The same postmarketing sections also list acute pancreatitis — which the trials did count, and where the numbers point in directions the warnings do not — ileus and intestinal obstruction, gallbladder disease including cholecystitis and cholelithiasis, acute kidney injury, and hypersensitivity reactions including anaphylaxis and angioedema. Gastroparesis and ileus are also the injuries at the centre of the federal GLP-1 lawsuits, where a 2025 ruling decided what counts as a gastroparesis diagnosis. Gallbladder disease is the one of those that is counted rather than merely listed, and the three weight-management labels disagree sharply about it — the stones and the weight loss run in opposite directions, which is not how the usual explanation goes. And in the Ozempic, Mounjaro and Saxenda labels, under skin and subcutaneous tissue disorders, alopecia — which is the subject of its own page here, because where that word appears and where it does not turns out to be the whole story. How these postmarketing entries sit beside the serious adverse events counted over up to seven years of trial follow-up is on GLP-1 long-term side effects.
What the reports database adds, and what it cannot
FDA's adverse event reporting system collects voluntary reports from patients, clinicians and manufacturers. It is not an incidence measure: there is no denominator, reports are unverified, a report is not a finding of causation, and the number of reports for a drug depends heavily on how long it has been marketed and how its manufacturer classifies what it receives. Nothing in it can be turned into a percentage.
Within those limits it does show what people and clinicians are moved to report. Queried on 2026-09-10, the most frequently reported terms for semaglutide were nausea (10,674), off-label use (7,325), vomiting (6,964), diarrhoea (6,462), decreased appetite (4,905) and constipation (4,724) — the label's own list, in roughly the label's own order, which is a mild sign the two sources are describing the same drug.
The eighth entry is the one worth flagging: impaired gastric emptying, 3,558 reports, above headache and fatigue. Delayed gastric emptying is the intended mechanism of this drug class rather than an incidental effect of it, which makes its appearance high on a spontaneous-report list a different kind of signal from nausea — it suggests a number of people are experiencing the mechanism itself as an adverse event. The labels' postmarketing sections do not carry a frequency for it either. For tirzepatide the label adds that this delay is largest after the first dose and fades.
For tirzepatide the same query returns something stranger, which is dealt with in full on our tirzepatide review: the most reported term is not a symptom at all.
The same database holds the answer to a question the labels refuse: what an overdose of these medicines looks like on the record. Not one of the seven approved labels states a dose that counts as one, and 95.6% of the tirzepatide reports carrying the term are logged non-serious.
Sources
- FDA label for Wegovy (semaglutide), Novo Nordisk, effective 2026-06-18, DailyMed, re-read 2026-09-23 (first read 2026-09-10 from a 2024 repackager copy with the same 2.4 mg figures).
- FDA label for Zepbound (tirzepatide) injection, effective 2026-08-28, same source and date.
- FDA labels for Ozempic (effective 2026-01-30), Mounjaro (effective 2026-08-27, Eli Lilly's copy, re-checked 2026-09-24) and Saxenda (effective 2026-02-25), same source and date, used for the postmarketing comparison.
- openFDA drug adverse event endpoint, counts by reaction term for semaglutide, queried 2026-09-10.