Insomnia on Zepbound and Other GLP-1 Drugs: What the Labels and Reports Record

Insomnia is barely in the GLP-1 labels. Nine current US prescribing labels were read for this page on 22 October 2026, and one of them — Saxenda — lists insomnia at all. Zepbound's label, the brand most searched alongside the word, has no insomnia line; what it does record about sleep comes from its sleep-apnea trials and points the other way. This page sets out the label record first, then the adverse-event reports and the record-based studies, with what each kind of evidence can and cannot show. The wider side-effect picture by brand is on GLP-1 side effects.

The label record: one entry in nine labels

Each label's sections 5, 6 and 14 searched for "insomnia" and "sleep", DailyMed, 22 October 2026.

Saxenda's entry is small. Table 2 lists adverse reactions reported by at least 2% of Saxenda-treated adults and more often than on placebo. Insomnia sits near the bottom: 2.4% of 3,384 people on Saxenda against 1.7% of 1,941 on placebo — about 81 people against 33, by our arithmetic from the percentages. Victoza, the same molecule at up to 1.8 mg for diabetes, does not list it; Saxenda goes to 3 mg. Whether the difference is dose, population (Saxenda's trials enrolled people for weight management) or chance, the label does not say.

An absence in a label is not proof of no effect. Labels list reactions that met a threshold (usually 2% or 5% and more often than placebo) in the trials behind them. A symptom reported less often, or equally often on placebo, does not appear. What the absence does show is that in trials of thousands of people, insomnia did not separate from placebo clearly enough to be tabled for eight of the nine products.

Zepbound: the sleep result the label does carry

Zepbound is the one GLP-1 drug with a sleep-related indication: moderate to severe obstructive sleep apnea in adults with obesity. Its label describes two 52-week trials (Study 5, 234 adults not using a breathing machine; Study 6, 235 adults on one) and reports that Zepbound-treated patients "showed improvement in sleep-related impairment compared to those who received placebo", measured with the PROMIS Sleep-Related Impairment 8a short form. Baseline daytime sleepiness on the Epworth scale averaged 10.5 and 10.

That is a measure of how much poor sleep interferes with daytime life in people with sleep apnea — not an insomnia scale, and not a population chosen for insomnia. It does mean the one sleep outcome measured in a Zepbound trial moved in the favourable direction. The apnea results themselves are on GLP-1 drugs and sleep apnea.

Adverse-event reports: present, uncounted as rates

FDA's Adverse Event Reporting System collects reports from patients, clinicians and manufacturers. openFDA's copy (data updated 30 July 2026) was queried on 22 October 2026 for reports naming each molecule and listing the MedDRA term "insomnia":

openFDA drug/event endpoint, patient.drug.openfda.generic_name and patient.reaction.reactionmeddrapt.exact:"INSOMNIA". The per-1,000 column is our division and describes reports, not patients.

What this table cannot do. A report says someone noticed a symptom while taking a drug, not that the drug caused it; one report can list many drugs and many symptoms; reporting rises with publicity and with how many people use a drug; and nobody knows how many people took each drug without reporting anything. The column is a share of reports, and it moves whenever the number of reports about other complaints moves, so a lower share for one molecule does not mean fewer people sleep badly on it. One 2025 analysis of these databases found a disproportionality signal for insomnia with dulaglutide in FAERS (reporting odds ratio 2.93, 95% CI 2.35–3.66) but not with the others (Katranski et al., International Journal of Clinical Pharmacy 2025). A signal of that kind marks something to study, not a measured risk.

Record-based studies: three answers, three populations

Large electronic-health-record studies compare people who started one drug with similar people who started another, and count new diagnoses. They are observational — people were not randomised — so their findings are associations.

Abstracts read on PubMed, 22 October 2026.

The tirzepatide–semaglutide difference in the 2026 study is a 13% relative difference in the rate of a recorded diagnosis, between two active drugs, in adults without diabetes. The same study found both drugs associated with lower hazards of anxiety and depression than naltrexone-bupropion. Neither study can separate a drug effect from differences in who gets prescribed which drug. How the class has been studied for mood is on GLP-1 drugs and depression.

How much research exists

A PubMed search on 22 October 2026 for the named GLP-1 drugs or the class name together with "insomnia" in the title or abstract returned 20 records. None is a trial designed to measure insomnia on a GLP-1 drug. They are record studies, adverse-event database analyses, a social-media analysis, commentary on sleep apnea and reviews. (Query: semaglutide, tirzepatide, liraglutide, dulaglutide or "GLP-1 receptor agonist(s)" [tiab] AND insomnia [tiab].)

What this page cannot tell you

• How often insomnia happens on Zepbound, Wegovy or Ozempic. No label reports a rate, and reports and record studies cannot supply one.
• Whether the time of day of the dose matters. The injectable labels reviewed (Zepbound, Mounjaro, Wegovy, Ozempic, Trulicity, Saxenda, Victoza) all allow the dose at any time of day; the semaglutide tablets are taken at least 30 minutes before the first food or drink of the day. None studies sleep by dosing time.
• Whether sleep troubles relate to the gastrointestinal effects. Nausea, reflux and fatigue are common early on and are recorded; no study counted here links them to insomnia.

Poor sleep has many causes, and a new or worsening sleep problem while starting or changing any medicine is worth raising with the prescriber, who can look at timing, dose steps and the person's other medicines.

Sources

• DailyMed prescribing information read 22 October 2026: Zepbound (set ID 487cd7e7-434c-4925-99fa-aa80b1cc776b, v40), Mounjaro (d2d7da5d-ad07-4228-955f-cf7e355c8cc0, v40), Wegovy (ee06186f-2aa3-4990-a760-757579d8f77b, v19), Ozempic (adec4fd2-6858-4c99-91d4-531f5f2a2d79, v20), Rybelsus/Ozempic tablets (27f15fac-7d98-4114-a2ec-92494a91da98, v14), Saxenda (3946d389-0926-4f77-a708-0acb8153b143, v22 — Table 2), Victoza (5a9ef4ea-c76a-4d34-a604-27c5b505f5a4, v31), Trulicity (463050bd-2b1c-40f5-b3c3-0a04bb433309, v62), Foundayo (8ac446c5-feba-474f-a103-23facb9b5c62, v10).
• openFDA drug adverse event API, data last updated 2026-07-30, queried 22 October 2026.
• Katranski J et al. Psychiatric adverse events linked to glucagon-like peptide 1 analogues. Int J Clin Pharm 2025;47(6):1739–1747. PMID 40522403.
• De Giorgi R et al. 12-month neurological and psychiatric outcomes of semaglutide use for type 2 diabetes. EClinicalMedicine 2024;74:102726. PMID 39764175.
• Huang YN et al. Neuropsychiatric association of tirzepatide and semaglutide in obesity with and without type 2 diabetes. Commun Med 2026;6:545. PMID 42373755.
• Liu TH et al. Psychiatric safety of GLP-1 and dual GIP/GLP-1 receptor agonists in adolescents with obesity. Diabetes Obes Metab, online 14 September 2026. PMID 42736027.
• PubMed search, 22 October 2026, query as stated above: 20 records.