What exists, and what does not
The gap on this topic is unusually clean. Migraine is one of the best-studied conditions in neurology, with trials that run to thousands of participants and a whole class of drugs developed in the last decade. Against that background, the cannabinoid evidence is startlingly thin.
- No randomised, placebo-controlled trial of CBD alone in migraine has been published.
- The trial evidence that exists concerns cannabis or THC-containing products, and is mostly observational.
- The largest quantitative dataset comes from an app, not a clinic.
The app study, and how to read it
In 2020, researchers at Washington State University published an analysis in The Journal of Pain of around 12,000 sessions logged in a cannabis tracking app by people using it for headache or migraine. Self-reported severity fell by roughly half after use, and men reported larger reductions than women.
The same analysis found no evidence that headaches got worse over time, but it did find that people used larger amounts as the months went on and that the size of the reported reduction fell. That pattern — needing more for less — is the one thing in this dataset that a person deciding whether to start should weigh most heavily.
Everything about the method makes it weaker than a trial: nobody was randomised, nobody was blinded, the ratings were entered by the same person who chose the product, and the products were overwhelmingly THC-containing flower rather than CBD. It tells you what people report. It cannot tell you what the drug did.
Why CBD gets discussed at all here
The reasoning usually runs through the endocannabinoid system: migraine involves the trigeminal system and CGRP, cannabinoid receptors are present in the relevant pathways, and a long-standing hypothesis called clinical endocannabinoid deficiency proposes that migraine, fibromyalgia and irritable bowel syndrome share an underlying cannabinoid tone problem.
It is an interesting hypothesis, first set out in 2004 and still discussed. It is a hypothesis. No trial has tested it in migraine and reported a result, and a mechanism that sounds plausible is a reason to run the study rather than a substitute for having run it.
The risk specific to this condition
Anything taken often for headache can produce medication overuse headache, in which the treatment becomes part of the cycle. It is well documented for over-the-counter painkillers and triptans, and it is the reason headache clinics ask how many days a month you take something.
Whether cannabinoids can do the same is not settled, but the pattern in the app data — using more over time for a smaller reported benefit — is the pattern that would produce it. Anyone taking anything for headache on more than ten days a month should be having that conversation with a clinician rather than with an article.
The symptom that appears when you stop
Cannabis withdrawal is a recognised syndrome, and headache is one of its documented features, alongside irritability, sleep difficulty and reduced appetite. It appears in people who have been using daily or near-daily, typically within the first week of stopping.
For someone using cannabis regularly for headaches, that produces a trap worth naming: headaches on the days without it, relief on the days with it, and a pattern that reads exactly like the product working. Frequency is the tell. If headaches cluster on days of lower use, that is the question to take to a clinician rather than an argument for using more.
This is not an argument that cannabis causes migraine. It is the reason the app study’s long-term finding — more used over time for a smaller reported benefit — deserves more weight than its headline number, and the reason a headache diary that records use as well as attacks answers a question that no amount of reading can.
What is worth doing instead
Migraine is one of the conditions where keeping a proper record changes the outcome, because attack frequency moves for reasons that are invisible without one.
| What to record | Why it earns its place |
|---|---|
| Attack days per month, not attacks | The number clinics use to decide whether preventive medication is warranted. |
| Days you took anything at all for it | The count that identifies medication overuse before it takes hold. |
| What you tried and at what point in the attack | Timing changes the outcome for most acute treatments, which makes "it did not work" ambiguous without it. |
| Sleep, cycle, alcohol, missed meals | Ordinary triggers that a month of records surfaces and memory does not. |
Four to eight weeks of that is worth more than anything on this page, and it is the record that makes a conversation with a neurologist productive rather than anecdotal.
What the evidence looks like elsewhere
Migraine is a useful place to see the difference between a thin evidence base and a thick one, because the thick one is right there for comparison. These are the things that have been through randomised trials for migraine, and what kind of role each has.
| Approach | Where it sits |
|---|---|
| Triptans | Taken during an attack. Three decades of randomised trials and the standard against which newer acute drugs are measured. |
| Gepants and ditans | Newer acute options for people who cannot take triptans, licensed on their own trial programmes. |
| CGRP monoclonal antibodies | Monthly or quarterly injections taken between attacks, developed specifically for migraine and trialled in thousands of people. |
| Topiramate, propranolol, amitriptyline | Older preventive medicines, borrowed from other fields, with long trial records in migraine. |
| Riboflavin, magnesium, feverfew | Modest evidence, graded as probably or possibly effective in headache society guidance, and the closest comparison to a supplement. |
| CBD alone | No randomised placebo-controlled trial published. |
The last row is the point of the table. This is not a condition where medicine has run out of options and a supplement is all that is left; it is a condition with an unusually active drug pipeline, where one particular product has skipped the step everything above it went through.
Reading a label without being taken in
If you are going to try CBD despite the state of the evidence, the label problem applies here as everywhere: a 2017 JAMA analysis of 84 extracts bought online found around seven in ten were inaccurately labelled, and some held undeclared THC.
| What to look for | Why it matters |
|---|---|
| Milligrams per serving, not per bottle | The large number on the front is usually the whole container. |
| Spectrum: full, broad or isolate | Full spectrum carries a trace of THC; broad spectrum removes it; isolate is CBD alone. |
| A batch certificate of analysis | Matching the batch code on your container and recently dated. |
| What else is in it | Feverfew, magnesium, riboflavin and caffeine all turn up in headache blends, and each has its own literature. |
That last row matters more here than on most topics, because several of the additives in headache products have been studied for migraine in their own right. A blend that seems to help tells you nothing about which ingredient did it.
Questions people ask
Has CBD been tested for migraine in a clinical trial?
Not on its own. No randomised, placebo-controlled trial of CBD alone in migraine has been published. The cannabinoid evidence that exists is largely observational and largely concerns THC-containing products.
What did the 12,000-session study show?
Self-reported headache and migraine severity fell by about half after use. It also showed people using larger amounts over time for smaller reported reductions, and nobody in it was randomised or blinded.
Can CBD make headaches worse?
Anything taken frequently for headache can feed medication overuse headache, which is well documented for painkillers and triptans. Whether cannabinoids do the same is unsettled, and it is a question for a clinician if you are taking something on more than ten days a month.
Is CBD or THC the one that has been studied?
Nearly all of it is THC-containing cannabis. That matters, because THC and CBD behave differently and a bottle of CBD is not a low-strength version of the thing that was studied.
Sources
- Cuttler C, et al. Short- and Long-Term Effects of Cannabis on Headache and Migraine. J Pain. 2020;21(5-6):722-730. accessed Sep 18, 2026
- Russo EB. Clinical endocannabinoid deficiency reconsidered. Cannabis Cannabinoid Res. 2016;1(1):154-165. accessed Sep 18, 2026
- Bonn-Miller MO, et al. Labeling Accuracy of Cannabidiol Extracts Sold Online. JAMA. 2017;318(17):1708-1709. accessed Sep 18, 2026
- Bonnet U, Preuss UW. The cannabis withdrawal syndrome: current insights. Subst Abuse Rehabil. 2017;8:9-37. accessed Sep 18, 2026
