Epilepsy holds the strongest evidence anywhere in cannabis medicine, and that evidence belongs to a prescription drug. Four randomized, placebo-controlled trials showed that purified CBD cut seizures in Dravet syndrome, Lennox-Gastaut syndrome and tuberous sclerosis complex. The product those trials tested is Epidiolex, a strawberry-flavored oral solution measured by syringe, dosed by the kilogram, and started only after a blood test of the liver.
Our position is short. There is no seizure-dose edible, and a CBD gummy bought for a seizure disorder is a different product wearing the same molecule's name. The arithmetic below shows why: to reach the dose the trials used, an adult would chew a bag and a half of gummies a day. The route to the evidence runs through a neurologist's prescription pad, and the antiseizure medication already in the cabinet stays where it is.
What does the evidence show for cannabis and seizures?
Strong results for one product in three syndromes, and very little for anything else. Purified CBD beat placebo in four randomized trials covering Dravet, Lennox-Gastaut and tuberous sclerosis complex, always added to existing antiseizure drugs. Dispensary oils, THC products and other epilepsy types have no comparable trials behind them.
| Trial | Syndrome and patients | Daily CBD dose | Result against placebo |
|---|---|---|---|
| Devinsky 2017, NEJM | Dravet; 120 children and young adults | 20 mg/kg | Monthly convulsive seizures fell from a median 12.4 to 5.9, against 14.9 to 14.1 on placebo |
| Devinsky 2018, NEJM (GWPCARE3) | Lennox-Gastaut; 225 patients aged 2 to 55 | 10 or 20 mg/kg | Drop seizures down 37.2% and 41.9%, against 17.2% on placebo |
| Thiele 2018, Lancet (GWPCARE4) | Lennox-Gastaut; 171 patients aged 2 to 55 | 20 mg/kg | Drop seizures down 43.9%, against 21.8% on placebo |
| Thiele 2021, JAMA Neurology (GWPCARE6) | Tuberous sclerosis complex; 224 patients aged 1 to 65 | 25 or 50 mg/kg | Seizures down 48.6% and 47.5%, against 26.5% on placebo |
Notice the placebo column. Seizure counts fell by 17% to 27% in the placebo arms of the two drop-seizure trials and the tuberous sclerosis trial, which is the number any parent's report of a homemade oil has to beat before it means anything. The Dravet trial is the sobering one on that front: 43% of the CBD group halved their convulsive seizures, and so did 27% of the placebo group.
Stockings and colleagues pooled the controlled and observational literature in 2018 for the Journal of Neurology, Neurosurgery and Psychiatry. Pharmaceutical CBD at 20 mg/kg a day made a 50% seizure reduction more likely than placebo, with eight patients treated for one extra responder, and it also raised the risk of serious adverse events. The randomized evidence, the authors noted, sits mostly in children with rare and severe syndromes, and trials of other syndromes and other cannabinoids were still missing.
What does the Epidiolex label say?
It approves Epidiolex for seizures of Lennox-Gastaut syndrome, Dravet syndrome and tuberous sclerosis complex in patients 1 year of age and older. Liver enzymes and bilirubin get checked before the first dose, and the label describes a titration that starts at 5 mg/kg a day and climbs toward 10 to 20 mg/kg, or 25 mg/kg in tuberous sclerosis.
The rest of the label reads like a list of reasons this belongs with a prescriber.
- Liver. ALT rose above three times the upper limit of normal in 13% of patients in the Lennox-Gastaut and Dravet trials, against 1% on placebo. With both valproate and clobazam on board, the rate was 30%; with neither, 3%.
- Clobazam. In healthy volunteers, adding CBD roughly tripled exposure to N-desmethylclobazam, clobazam's active metabolite. The label suggests considering a lower clobazam dose if clobazam side effects appear.
- Sleepiness. In the Lennox-Gastaut and Dravet trials, somnolence and sedation hit 32% of treated patients, and 46% of those also taking clobazam.
- Food. In healthy volunteers, a high-fat meal raised peak CBD levels five-fold. The label asks for consistent dosing with respect to meals, which is the opposite of how anyone eats a gummy.
- Stopping. The dose comes down gradually, because abrupt discontinuation of an antiseizure drug risks more seizures and status epilepticus.
One more line in section 9 surprises people: Epidiolex is not a controlled substance. The prescription exists because of the dosing, the liver monitoring and the interactions, and none of those disappear when the CBD comes from a hemp store instead.
How many CBD gummies would match an Epidiolex dose?
Twenty-eight a day at the low end for a 70kg adult, and 56 at the top Lennox-Gastaut and Dravet dose. The trials used 10 to 20 mg/kg a day, which is 700 to 1,400mg of CBD for that adult, and a typical retail gummy carries 25mg. Tuberous sclerosis dosing at 25 mg/kg works out to 1,750mg, or 70 gummies.
We ran the numbers on a real bag. Medterra's Calm CBD Gummies list 25mg of broad-spectrum CBD and 125mg of L-theanine per piece, at $29.99 for 20 on the brand site on October 27, 2026. At 10 mg/kg, the 70kg adult finishes the bag before dinner and still needs eight more, spends about $42 a day, and swallows 3,500mg of L-theanine along the way. At 20 mg/kg it is $84 a day and 56 pieces of strawberry lemonade. For a 20kg child, the trial range comes to 8 to 16 gummies daily. Epidiolex delivers 100mg in a single millilitre.
Nobody in a real kitchen does this, and that is the point. What families give at home sits far below the trial range. In Knupp's 2019 prospective study of children starting oral cannabis extracts, the median highest CBD dose reached was 0.9 mg/kg a day, about a tenth of the lowest dose that beat placebo.
Can you trust the CBD number on the label?
Less than a third of the time, in the best-known test. Bonn-Miller's team bought 84 CBD extracts online from 31 companies in 2016 and had them analyzed blind: 31% matched their label within 10%, 43% held more CBD than stated, 26% held less, and 21% contained THC the label did not list.
The FDA's own testing found worse. Its 2015 results list several products sold as CBD that came back negative for cannabinoids altogether, including a capsule bottle labeled 500mg of CBD. Among its 2016 results is a product sold on Etsy as a 25mg-per-capsule CBD oil, which tested at 0.031mg of CBD per gram and 17mg of delta-9 THC per gram. That is a THC capsule with a CBD label, and THC is the cannabinoid with the murkier relationship to seizures. The American Epilepsy Society cites exactly this problem: no USP standard exists for cannabis products, and a dispensary or over-the-counter CBD product can carry THC, pesticides and other impurities in unknown amounts.
Does THC help seizures or trigger them?
Both are reported, and neither is settled. Surveys at epilepsy centers find many patients who use cannabis say it helps, and a minority who say it set off seizures. Synthetic cannabinoids and concentrated THC have case reports of seizures. No controlled trial has tested THC for epilepsy.
The survey numbers are worth reading closely. In a 2004 telephone survey at a tertiary epilepsy center, 21% of patients had used marijuana in the past year, and most active users reported a benefit. At a Canadian epilepsy monitoring unit, 84% of cannabis users with epilepsy said their seizures improved, and so did 72.7% of users whose events turned out to be psychogenic nonepileptic seizures, which are not epileptic at all. Perceived seizure benefit tracked perceived stress relief. Five patients with epilepsy in that unit described cannabis as a possible seizure trigger. A 2014 interview study at another tertiary epilepsy clinic found 84.1% of cannabis users reported no effect on their epilepsy either way.
On the other side, synthetic cannabinoids carry reports of seizures, coma and posterior reversible encephalopathy syndrome, per a 2024 Seminars in Neurology review. Kahan's 2021 case report describes a 17-year-old with no seizure history who had a generalized tonic-clonic seizure while vaping concentrated THC wax daily; her abnormal EEG had resolved a week later without antiseizure medication. The AES summary is that scientific evidence for cannabis itself in epilepsy is highly limited, and the absence of data leaves clinicians unable to make rational decisions about it.
Is cannabis oil an option for seizures in kids?
The evidence-based option is Epidiolex, prescribed by a neurologist for children 1 year and older with Dravet, Lennox-Gastaut or tuberous sclerosis complex. Artisanal oils have been studied mostly through parent reports in Colorado, where response rates looked like placebo once measured prospectively, and some children had more seizures.
Press, Knupp and Chapman at Children's Hospital Colorado reviewed 75 children given oral cannabis extracts and published the results in Epilepsy and Behavior in 2015. Parents reported a 50% seizure reduction in 33%. The responder rate was 47% for families who had moved to Colorado for the oil and 22% for families already living there, which says a lot about expectation. EEGs in responders did not improve. Adverse events hit 44%, including more seizures in 13%, and the rare events list ran to status epilepticus requiring intubation, and death.
A 2017 follow-up of 119 children found 71% had stopped their extract during the study period, and average use lasted 11.7 months. Then the same research group did it prospectively, with seizure diaries and blood levels: 5 of 21 children (24%) responded, which the authors themselves compared to placebo response in the pharmaceutical trials, and 3 stopped early for a perceived increase in seizures. Parents searching at midnight deserve that answer plainly. A pediatric epilepsy specialist can prescribe the version that worked in trials, with the liver tests and interaction checks that come attached.
Why does the antiseizure medication stay put?
Because CBD was tested as an add-on, and because the Epidiolex label warns that stopping antiseizure drugs abruptly risks status epilepticus, a prolonged seizure emergency. The AES position is that RCT evidence is lacking for CBD as a monotherapy for any seizure type. Every trial in the table kept patients on their existing drugs.
The interactions are measured. In Geffrey's 2015 study of 13 children on clobazam starting CBD, levels of the active clobazam metabolite rose 500% on average by week four, and 10 of the 13 had their clobazam dose reduced. Gaston's 2017 Alabama program found CBD shifted serum levels of clobazam, rufinamide, topiramate, zonisamide and eslicarbazepine, and higher liver enzymes in people also taking valproate. Both studies gave pharmaceutical CBD through expanded-access programs, and nobody has measured what a nightly 25mg gummy does to a clobazam level. That unknown is the reason to tell the neurologist, who can order the blood test.
If the interest is in cannabis for a different symptom that comes with epilepsy, the evidence for those questions lives on separate pages, such as our THC and CBD comparison, and the seizure plan does not change to make room for it.