The link between body weight and migraine is real, and it is unusually specific. A meta-analysis of 12 studies covering 288,981 participants found the age- and sex-adjusted risk of migraine raised 27% in people with obesity against normal weight (OR 1.27; 95% CI, 1.16 to 1.37; p < 0.001), an association that survived multivariate adjustment. In overweight participants the association was smaller (OR 1.08) and lost significance after adjustment. Underweight participants carried a 13% increase that did survive it, making the curve a U rather than a slope.[1]
A telephone survey of 30,215 people sharpened the point two decades earlier. Transformed migraine ran from 0.9% at normal weight to 1.2% in the overweight (OR 1.4), 1.6% in the obese (OR 1.7) and 2.5% in the morbidly obese (OR 2.2; 95% CI, 1.5 to 3.2). Chronic tension-type headache showed an effect only in the morbidly obese group, and in adjusted analysis obesity was associated with migraine but not with tension-type headache at all.[2] The weight association belongs to migraine specifically. Ordinary headache on the drug is a different question, answered in the headache article.
None of that establishes that removing the weight removes the migraine, and the literature that would establish it does not exist.
The intracranial-pressure hypothesis, and where it came from
The mechanism under investigation does not run through the migraine pathway at all. It runs through cerebrospinal fluid. GLP-1 receptor agonism reduces intracranial pressure, and the drug class is being trialed in idiopathic intracranial hypertension — a disorder of raised pressure with papilledema, in which headache is a symptom. A randomized, placebo-controlled trial using telemetric intracranial pressure catheters in women with active disease found exenatide lowered pressure at 2.5 hours, 24 hours and 12 weeks; fifteen of sixteen recruited participants completed it.[3] Those results, and the headache-day reductions that followed them, are covered in the headache article and in the eye-health article, because the visual endpoints are the reason the trial exists.
The step from there to migraine has been taken once, in rats. A 2025 rat study raised intracranial pressure by intracisternal kaolin injection — median 15.96 mmHg against 6.02 in controls (P = 0.0007) — and measured the consequences. Periorbital mechanical withdrawal thresholds fell from 6.13 to 2.35 g over seven days (P < 0.001), spreading depolarization lengthened, cerebral blood flow response dropped, and ADP, ATP and NAD⁺ were depleted. Exenatide lowered pressure to 9.74 mmHg against 18.27 on vehicle (P = 0.004) and rescued both the withdrawal thresholds and the depolarization response.[4]
The part of that rat experiment that matters most is the arm nobody quotes. The CGRP receptor antagonist olcegepant — a migraine drug, not a pressure drug — also prevented the periorbital threshold change. The authors’ own conclusion is that calcitonin gene-related peptide drives the pain in raised pressure, and that blocking CGRP is the therapeutic implication. The GLP-1 agonist was acting upstream on the pressure, not on the migraine pathway.
The only other mechanistic work is a 2023 mouse study in which liraglutide attenuated trigeminal allodynia in a nitroglycerin-induced chronic migraine model by stimulating IL-10 release in the trigeminal nucleus caudalis.[5] A mouse and a rat are not a migraine patient, and both models induce the condition artificially.
The human experiment that constrains all of this
Migraine research has a standard way of asking whether a circulating substance acts on the trigeminovascular system: infuse it and see. That experiment has been done with GLP-1. In a crossover, double-blind, placebo-controlled randomized design, 21 healthy volunteers received intravenous GLP-1 at 2.5 pmol/kg/min or placebo over 20 minutes on two days at least a week apart, with transcranial Doppler and ultrasound dermascan measurement. Twenty were analyzed.
Nothing moved. Mean blood flow velocity in the middle cerebral artery was unchanged (P = 0.227), as were superficial temporal artery diameter (P = 0.096), radial artery diameter (P = 0.221) and facial skin blood flow (P = 0.814). Headache occurred in 10 participants after GLP-1 and 7 after placebo(P = 0.250).[6]
Read in both directions, that is the most useful result on this page. GLP-1 is not a migraine trigger: an infusion that reliably provokes attacks when the substance is CGRP or PACAP provoked nothing here. And GLP-1 is not doing anything to cranial arteries in a person with normal pressure, which means the pressure route is the only live mechanism and it requires raised pressure to start with.
The treatment evidence, counted
As of 15 September 2026, PubMed holds one record typed as a randomized controlled trial and carrying migraine together with any of GLP-1, semaglutide, liraglutide, tirzepatide, exenatide or dulaglutide in its title or abstract — and that record is the infusion study above, which is not a treatment trial. The identical query shape holds 326 records for the CGRP-targeting migraine drugs and 99 for topiramate, and substituting idiopathic intracranial hypertension for migraine holds 2. The query is not the reason the count is one. There is no randomized trial of this class as a migraine treatment.
What exists instead is one prospective open-label pilot. Thirty-one patients with obesity and high-frequency or chronic migraine, unresponsive to at least two preventives, took liraglutide 1.2 mg daily for 12 weeks. Monthly headache days fell from 19.8 to 10.7, a mean difference of 9.1 days (95% CI, 5.41 to 12.84; p < 0.001; Cohen’s d 0.90), while BMI moved from 34.0 to 33.9 and the change in BMI did not predict the change in headache frequency.[7] A nine-day reduction is a large effect. It also has no placebo arm, in a disorder whose placebo response in randomized preventive trials is among the largest in medicine, and 31 people is a pilot.
Below that sit two patients. A 2026 case series describes two people whose migraine frequency fell after starting tirzepatide; its authors name the accompanying weight loss as a significant confounder and ask that the cases be read as preliminary observations.[8] A broader systematic review of this class across headache and pain disorders screened 833 records and included 42 studies spanning inflammatory, neuropathic, visceral and headache pain, which is the size of the evidence base across every pain domain combined.[9]
And one case runs the other way. A man in his early 50s with a history of hemiplegic migraine once or twice a year reported daily migraines, many of them hemiplegic, across 60 days of liraglutide for weight loss; they resolved only on discontinuation.[10] One case report is not a rate. It is enough to make “this class treats migraine” an overstatement of a literature that also contains an exacerbation.
The one measurement at population scale, and its subgroup
A 2026 nationwide interrupted time series used Danish health registers to test whether starting semaglutide for weight management changed triptan consumption. All 189,392 adults initiating semaglutide between December 2022 and December 2024 were included — 68% female, median age 50 — with a 24-month baseline and 12-month follow-up. Triptan use had been rising before initiation; afterward the trend reversed, falling 13 defined daily doses per month per 10,000 people (95% CI, −25 to −1.3), a 7% relative reduction at 12 months (RR 0.93; 0.88 to 0.97). The decrease came from lower consumption among existing triptan users (RR 0.86; 0.82 to 0.90) rather than from fewer people starting them, and was largest in 18- to 35-year-olds (RR 0.86; 0.78 to 0.94) and in prior users of preventive medication (RR 0.88; 0.82 to 0.94).[11]
Then the subgroup that changes the sentence. Women showed an 8% reduction (RR 0.92; 0.88 to 0.97). Men showed no statistically significant change at all.[11] In the largest measurement anyone has made, the effect is present in one sex and absent in the other — which is also where the weight-loss response differs, as set out in the men’s health article.
A 7% fall in triptan defined daily doses is also not a 7% fall in migraine. The design cannot separate the drug from the weight it removed, and it cannot exclude the simplest confounder available: a person in the first months of titration is frequently nauseated, and swallowing a triptan is one of the things nausea discourages. Those months are described in the side-effect article.
What is extrapolation, stated as such
Four things on this page are borrowed rather than measured in migraine. The intracranial pressure results come from a different disease, in women with papilledema. The allodynia and neurovascular results come from rats and mice with artificially induced conditions. The headache-day reductions in that pressure literature were counted in patients whose headache had a mechanical cause a migraine patient does not have. And the triptan series measures a prescription record, not an attack.
One thing is measured directly in humans and it is a negative: infused GLP-1 produced no cranial vasodilation and no excess headache. One thing is measured directly in migraine patients and it is uncontrolled: 31 people, open label, nine fewer headache days.
That is the whole file. It is enough to justify a randomized trial and not enough to justify a purchase, and no regulator has approved any drug in this class for migraine.
What a migraine patient should actually do with this
If a GLP-1 is already indicated for weight, migraine is not a reason to avoid it: the provocation study argues against a triggering effect, and the only population-scale signal points, weakly, the helpful way. Existing preventive and acute treatment should not be changed on the strength of any of the above, and a prescriber managing migraine is the person to tell about a new weekly injection — not least because appetite suppression, skipped meals and dehydration are established attack triggers in their own right.
Everything above was measured with FDA-approved product at labeled doses. Most sellers reviewed here dispense compounded semaglutide and tirzepatide, which are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before they are dispensed, and no compounded product has been studied in a headache disorder at all. Anyone starting one for weight will find the field and its prices on the compounded semaglutide board.