GLP-1 side effects: incidence, and the timeline that is not published

Start with the first table, which answers the question the page is usually asked. Every label states that the gastrointestinal reactions cluster during dose escalation and every label defines that phase in weeks or days, so the phase column is real and sourced. The onset and resolution columns read "Not stated" all the way down because that is what five full labels say. If you have seen a chart claiming nausea peaks in week three and clears by week eight, it was not built from a label.

In every incidence table, read the placebo column first. A label table prints what happened on the drug, not what the drug caused, and in these trials the placebo arms are large: 16% of people on placebo in the Wegovy trials reported nausea. The "above placebo" column is that subtraction, and the last column turns it into the sentence a reader wants — how many people take the drug for one of them to get the reaction who would not have had it anyway.

Rows are ordered by attributable excess, not by the raw percentage, so the order differs from the label's. That reordering is the point: on Wegovy, diarrhea has the bigger headline number and vomiting is the bigger drug effect.

Never compare a figure in one table against a figure in another. Each is a separate pooled analysis with its own population, exposure and reporting threshold — the two diabetes tables print reactions at 5% and above, the three weight-management tables at 2% and above, so a reaction missing from a diabetes table may simply have fallen below a higher bar. The column heads carry the arm sizes so you can see what each table is actually made of.

What the labels say about WHEN — and what they do not

One row per approved product. The escalation phase is computed from each label’s own dosing section; the last two columns are what five full prescribing informations contain on timing.

The two tirzepatide labels are the only ones that claim improvement, and they claim it in four words. Mounjaro and Zepbound, section 6.1, verbatim: "The majority of nausea, vomiting, and/or diarrhea events occurred during dose escalation and decreased over time." Ozempic, section 6.1, verbatim: "The majority of reports of nausea, vomiting, and/or diarrhea occurred during dose escalation." — and it stops there. Wegovy, section 6.1, verbatim: "These reactions were most frequently reported during dosage escalation." Wegovy TABLETS have neither a sentence of their own nor an incidence table; the escalation sentence in that label sits inside a paragraph reporting the injection's figures. Every one of the seven does state in section 2 why the ladder exists — Zepbound, verbatim: "Follow the dosage escalation below for all indications to reduce the risk of gastrointestinal adverse reactions" — which is a weaker claim than the 6.1 sentence and is the only thing the Wegovy tablet row can stand on. One label states when people who stop tend to stop, and only one: Zepbound, section 6.1, verbatim: "The majority of patients who discontinued ZEPBOUND due to adverse reactions did so during the first few months of treatment due to gastrointestinal adverse reactions.". "A few months" is the most precise timing statement in the entire corpus.

Wegovy injection 2.4 mg — weight reduction in adults

Reported by at least 2% of patients and more often than placebo, across three placebo-controlled trials. Ordered by how much the drug adds.

Label table title, verbatim: "Table 3. Adverse Reactions (≥2% and Greater Than Placebo) in WEGOVY 2.4 mg Injection-treated Adults with Obesity or Overweight for Weight Reduction". Exposure: 2,116 adults treated for up to 68 weeks, plus a 7-week off-drug follow-up. Across those trials 73% of patients on Wegovy injection and 47% on placebo reported any gastrointestinal adverse reaction — an excess of +26 pts, so roughly one in 4 — and severe gastrointestinal reactions were reported in 4.1% versus 0.9%. The eructation row's placebo figure is printed in the label as "<1%", so its excess is a floor rather than an exact figure. The label states no time to onset and no duration for any row in this table.

Wegovy HD 7.2 mg — the reactions that separate it from 2.4 mg

A different table with a different bar: reactions at 2% or more AND more common than on 2.4 mg as well as placebo. Two 72-week trials.

Label table title, verbatim: "Table 4. Adverse Reactions (2% and Greater Than WEGOVY 2.4 mg and Placebo) in WEGOVY 7.2 mg Injection-treated Adults with Obesity for Weight Reduction". This table exists to show what the extra 4.8 mg buys and costs — a dose three times the 2.4 mg maintenance — and one row dominates it: dysesthesia — the label's composite for altered skin sensation — at 22% on 7.2 mg against 6% on 2.4 mg and 0% on placebo. No risk ratio is printed for it because the placebo arm is zero and a ratio against zero is not a number. Section 6.1 is also the only place in any of these five labels that describes an event resolving, and it does so without a figure: "When an action was taken with WEGOVY, events resolved faster than the events where no action was taken with WEGOVY", alongside "Among patients who experienced dysesthesia with WEGOVY 7.2 mg injection, 18% did not report recovering during the trial duration". Of 38 patients who recovered and were re-escalated to 7.2 mg, 17 of them — 45% — reported a recurrence.

Zepbound injection — weight reduction and long-term maintenance

Three maintenance doses side by side, from two placebo-controlled trials running up to 72 weeks. Ordered by how much the drug adds at its worst dose.

Label table title, verbatim: "Table 1: Adverse Reactions (≥2% and Greater than Placebo) in ZEPBOUND-Treated Adults with Obesity or Overweight in Weight Reduction and Long-term Maintenance Trials (Study 1 and Study 2)". Exposure: 2,519 patients across two trials, up to 72 weeks plus 4 weeks off drug. Any gastrointestinal adverse reaction was reported by 56%, 56%, 56% at 5, 10 and 15 mg against 30% on placebo — note that the three dose arms are identical, which is the clearest single indication in this corpus that the dose you settle on is not what determines whether you get gastrointestinal reactions. Two rows do not rise monotonically with dose: constipation runs 17%, 14%, 11% and dizziness runs 4%, 5%, 4%, both highest at a middle dose. Hypotension's placebo arm is 0%, so no ratio is printed for it.

Mounjaro injection — type 2 diabetes

A different population and a higher bar: reactions reported by at least 5% of patients across two placebo-controlled diabetes trials.

Label table title, verbatim: "Table 1: Adverse Reactions in Pool of Placebo-Controlled Trials Reported in ≥5% of MOUNJARO-treated Adult Patients with Type 2 Diabetes Mellitus", with the label's own note that "Percentages reflect the number of patients who reported at least 1 occurrence of the adverse reaction". Exposure: 718 patients, mean duration of exposure 36.6 weeks. These figures are far below Zepbound's for the same molecule at the same doses — nausea 18% here against 28% there at 15 mg — and the reason is not that Mounjaro is gentler. It is a different population, a shorter mean exposure (718 patients, mean duration of exposure 36.6 weeks against up to 72 weeks) and a 5% reporting bar instead of 2%. The two tables must not be read against each other, and this footnote exists because the temptation to do so is the whole reason people get this wrong.

Ozempic injection — type 2 diabetes

The smallest pool here and the only table whose figures the label prints to one decimal place. Reactions reported by at least 5% of patients.

Label table title, verbatim: "Table 1. Adverse Reactions in Placebo-Controlled Trials Reported in ≥5% of OZEMPIC-Treated Patients with Type 2 Diabetes Mellitus". Exposure: 521 patients, mean duration of exposure 32.9 weeks. Two rows fall as the dose rises — abdominal pain 7.3% then 5.7% and constipation 5% then 3.1% — which is what small arms of about 260 patients look like, not a protective effect of a higher dose. Below the 5% bar the label separately lists dyspepsia (1.9%, 3.5%, 2.7%), eructation (0%, 2.7%, 1.1%), flatulence (0.8%, 0.4%, 1.5%), gastroesophageal reflux disease (0%, 1.9%, 1.5%) and gastritis (0.8%, 0.8%, 0.4%), stated for placebo, 0.5 mg and 1 mg respectively. Any gastrointestinal reaction: 32.7% and 36.4% against 15.3% on placebo.

Semaglutide tablets — type 2 diabetes

One pooled analysis covering both Rybelsus and Ozempic tablets, which the label treats as a single product throughout this section.

Label table title, verbatim: "Table 2. Adverse Reactions in Placebo-Controlled Trials Reported in ≥5% of Semaglutide Tablets-Treated Patients with Type 2 Diabetes Mellitus". Exposure: 1,071 patients, mean duration of exposure 41.8 weeks. The label reports these as "semaglutide tablets 7 mg" and "14 mg" and does not separate Rybelsus from Ozempic tablets anywhere in section 6.1, which is consistent with section 2.2 pairing Rybelsus 7 mg with Ozempic tablets 4 mg and Rybelsus 14 mg with Ozempic tablets 9 mg. Any gastrointestinal reaction: 41% at 14 mg, 32% at 7 mg, 21% on placebo, including severe reactions at 2%, 0.6% and 0.3%. There is no incidence table anywhere for WEGOVY 25 mg tablets — see the last table on this sheet.

How many people stop, and how many stop because of the gut

The figure that matters more than any single reaction rate. Each row is its own pooled analysis; read down a row, never across two.

The Wegovy tablet row has no incidence figures because the label publishes none. Section 6.1, verbatim: "In the clinical trial with WEGOVY 25 mg tablets, the types and frequency of common adverse reactions were similar to those listed in Table 3." That is an assertion of similarity, not a dataset, and printing Table 3's percentages under a tablet heading would be inventing a trial result. What the tablet does have is its discontinuation rates, and they are here. Note also that the all-cause column is blank for three of the seven rows: Ozempic, Mounjaro and the semaglutide tablets report a gastrointestinal discontinuation rate in section 6.1 and no overall figure alongside it, so none is printed. Zepbound is the only product whose label says WHEN people stopped: "The majority of patients who discontinued ZEPBOUND due to adverse reactions did so during the first few months of treatment due to gastrointestinal adverse reactions."

Frequently asked questions

How long do GLP-1 side effects last?

No FDA label answers this, and any page giving you a number in weeks did not get it from one. All five current labels — Ozempic, Wegovy, the oral semaglutide tablets, Mounjaro and Zepbound — were read in full for this chart and none states a time to onset, a median duration or a time to resolution for any adverse reaction. What they do state is the phase: the gastrointestinal reactions cluster during dose escalation, which is weeks 1–16 on Wegovy injection and weeks 1–20 on Zepbound. Two of the labels add that the reactions "decreased over time" — both tirzepatide products — and the three semaglutide labels do not say it.

How common is nausea on Wegovy?

44% of adults on Wegovy injection 2.4 mg reported nausea across three placebo-controlled trials, against 16% of those on placebo. The difference is +28 pts, which means roughly 1 person in 4 gets nausea because of the drug — the rest would have reported it either way. Any gastrointestinal reaction at all was reported by 73% on Wegovy and 47% on placebo, and severe gastrointestinal reactions by 4.1% against 0.9%.

Do GLP-1 side effects get better over time?

The two tirzepatide labels say so and the three semaglutide labels do not, which is a real difference between documents rather than between drugs. Mounjaro and Zepbound, section 6.1, verbatim: "The majority of nausea, vomiting, and/or diarrhea events occurred during dose escalation and decreased over time." Ozempic and the tablets say the reactions "occurred during dose escalation" and stop; Wegovy says they "were most frequently reported during dosage escalation" and stops. The published trials fill part of the gap — STEP 1 reports that nausea and diarrhea "were typically transient and mild-to-moderate in severity and subsided with time" — but that is a journal, not the approved labelling, and it too attaches no interval.

Which GLP-1 has the fewest side effects?

This chart will not answer that, and neither can the labels. Each table is a separate pooled analysis with its own population, its own exposure and its own reporting bar: the weight-management tables report reactions at 2% and above in people with obesity treated for up to 72 weeks, the diabetes tables at 5% and above in people with type 2 diabetes treated for a mean of 33 to 42 weeks. Mounjaro's nausea figure of 18% at 15 mg sits beside Zepbound's 28% at 15 mg for the identical molecule at the identical dose — the gap is the trial design, not the drug. Only a head-to-head trial can rank two products, and none of these tables is one.

How many people stop taking a GLP-1 because of side effects?

Fewer than most people expect. Permanent discontinuation for a gastrointestinal adverse reaction was 4.3% on Wegovy injection 2.4 mg against 0.7% on placebo, 4.3% on Zepbound 15 mg against 0.5%, and 6.6% on Mounjaro 15 mg against 0.4%. Discontinuation for any reason where the label states it was 6.8% on Wegovy and 6.7% on Zepbound 15 mg. One label says when: Zepbound's, which states the majority of those who stopped "did so during the first few months of treatment".

Why does the Wegovy 7.2 mg table look so different?

Because it is built on a different question. Table 4 lists only reactions at 2% or more that were also more common on 7.2 mg than on 2.4 mg AND placebo, so it shows what the extra dose adds rather than what the drug does. One row dominates it: dysesthesia — the label's composite for altered skin sensation, including paresthesia, burning and sensitive skin — at 22% on 7.2 mg, 6% on 2.4 mg and 0% on placebo. It is also the only adverse reaction in any of these five labels for which recovery is described at all, and the description is not a timeline: 18% of affected patients "did not report recovering during the trial duration", and of 38 who recovered and returned to 7.2 mg, 17 had it again.

Does the incidence table cover Wegovy tablets?

No, because the label does not publish one. Section 6.1 describes the tablet trial — 204 adults treated for up to 64 weeks — reports that 6.9% discontinued for an adverse reaction against 5.9% on placebo and 3.4% for a gastrointestinal one against 2%, and then says that "the types and frequency of common adverse reactions were similar to those listed in Table 3", which is the injection table. An assertion of similarity is not a dataset. Copying the injection percentages into a row headed Wegovy tablets would be publishing a trial result nobody published, so this chart carries the figures the tablet actually has and leaves the rest blank.

Are the side effects worse on a higher dose?

Less consistently than the dose-response story suggests. Nausea on Zepbound runs 25%, 29%, 28% across 5, 10 and 15 mg — flat, and highest at the middle dose. Constipation runs 17%, 14%, 11%, falling as the dose rises. Any gastrointestinal reaction is 56%, 56%, 56% — identical at all three. Vomiting does rise with dose, 8%, 11%, 13%, and so does discontinuation. The labels' own explanation is that the reactions belong to the escalation rather than to the destination, which is why every one of them tells you to climb slowly rather than to stop low.

Sources

1. [1]WEGOVY (semaglutide) injection and tablets — full US prescribing information, Novo Nordisk. Section 6.1 Table 3 (2.4 mg injection), Table 4 (7.2 mg injection), the WEGOVY 25 mg tablet paragraph, gastrointestinal adverse reactions and dysesthesia. DailyMed set id ee06186f-2aa3-4990-a760-757579d8f77b
2. [2]ZEPBOUND (tirzepatide) injection — full US prescribing information, Eli Lilly and Company. Section 6.1 Table 1 and the gastrointestinal adverse reactions and discontinuation paragraphs; section 2.1 dose escalation. DailyMed set id 487cd7e7-434c-4925-99fa-aa80b1cc776b
3. [3]MOUNJARO (tirzepatide) injection — full US prescribing information, Eli Lilly and Company. Section 6.1 Table 1 and gastrointestinal adverse reactions; section 2.1 dose escalation. DailyMed set id d2d7da5d-ad07-4228-955f-cf7e355c8cc0
4. [4]OZEMPIC (semaglutide) injection — full US prescribing information, Novo Nordisk. Section 6.1 Table 1, the sub-5% gastrointestinal reactions and the discontinuation figures; section 2.2 dose escalation. DailyMed set id adec4fd2-6858-4c99-91d4-531f5f2a2d79
5. [5]RYBELSUS and OZEMPIC (oral semaglutide) tablets — full US prescribing information, Novo Nordisk. Section 6.1 Table 2 and gastrointestinal adverse reactions; section 2.2 dose escalation. DailyMed set id 27f15fac-7d98-4114-a2ec-92494a91da98
6. [6]Once-Weekly Semaglutide in Adults with Overweight or Obesity. New England Journal of Medicine, 2021 (STEP 1). PMID 33567185 — quoted for what it says about the time course of adverse events, and noted for what it still does not supply
7. [7]Gastrointestinal tolerability of once-weekly semaglutide 2.4 mg in adults with overweight or obesity, and the relationship between gastrointestinal adverse events and weight loss. Diabetes, Obesity and Metabolism, 2022 (pooled STEP 1–3). PMID 34514682 — quoted for what it says about the time course of adverse events, and noted for what it still does not supply
8. [8]Tirzepatide Once Weekly for the Treatment of Obesity. New England Journal of Medicine, 2022 (SURMOUNT-1). PMID 35658024 — quoted for what it says about the time course of adverse events, and noted for what it still does not supply
9. [9]Gastrointestinal tolerability and weight reduction associated with tirzepatide in adults with obesity or overweight with and without type 2 diabetes in the SURMOUNT-1 to -4 trials. Diabetes, Obesity and Metabolism, 2025. PMID 39789843 — quoted for what it says about the time course of adverse events, and noted for what it still does not supply
10. [10]PIONEER 1: Randomized Clinical Trial of the Efficacy and Safety of Oral Semaglutide Monotherapy in Comparison With Placebo in Patients With Type 2 Diabetes. Diabetes Care, 2019. PMID 31186300 — quoted for what it says about the time course of adverse events, and noted for what it still does not supply