Yes, randomized clinical trials show that a cannabis compound treats certain types of epilepsy: four phase 3 trials found that prescription cannabidiol (CBD), sold as Epidiolex, cut seizures by roughly 40% to 50% in Dravet syndrome, Lennox-Gastaut syndrome and tuberous sclerosis complex, compared with about 15% to 27% on placebo. Those trials led to FDA approval. They tested purified CBD added to standard seizure drugs, not marijuana, and they do not show benefit for other types of epilepsy.
Headlines saying “study shows cannabis treats epilepsy” have run for decades, often based on one or two patients. This article focuses on the studies that actually settled the question: the placebo-controlled CBD trials published between 2017 and 2021. It walks through each trial, what was measured, the side effects, and the limits. For the broader picture, including early research, THC and the legal status of cannabis for seizures, see our main guide to cannabis and epilepsy research.
The Epidiolex Trials at a Glance
| Trial | Condition and patients | CBD dose | Result (median seizure reduction) |
|---|---|---|---|
| Devinsky et al., NEJM 2017 | Dravet syndrome, 120 children and young adults | 20 mg/kg/day | Convulsive seizures 12.4 to 5.9 a month on CBD vs 14.9 to 14.1 on placebo |
| GWPCARE3, NEJM 2018 | Lennox-Gastaut syndrome, 225 patients | 10 or 20 mg/kg/day | Drop seizures down 37.2% (10 mg) and 41.9% (20 mg) vs 17.2% on placebo |
| GWPCARE4, Lancet 2018 | Lennox-Gastaut syndrome, 171 patients aged 2 to 55 | 20 mg/kg/day | Drop seizures down 43.9% vs 21.8% on placebo |
| GWPCARE6, JAMA Neurology 2021 | Tuberous sclerosis complex, 224 patients aged 1 to 65 | 25 or 50 mg/kg/day | Seizures down 48.6% (25 mg) and 47.5% (50 mg) vs 26.5% on placebo |
In every trial, patients kept taking their existing seizure medicines. CBD was tested as an add-on.
Why These Studies Count as Strong Evidence
Before 2017, the evidence was thin. A 2014 Cochrane review found only four small, low-quality randomized reports covering 48 people and said no reliable conclusions were possible. The newer trials fixed the main problems:
- Randomized and double-blind: neither families nor investigators knew who got CBD and who got a matching placebo.
- Placebo-controlled: seizure counts in rare epilepsies swing a lot, and caregivers’ hopes can color reports. The placebo groups improved too, by 15% to 27%, which shows why a comparison group is essential.
- Pre-specified outcomes: each trial named its main measure, such as monthly drop seizures, in advance.
- Replication: the Lennox-Gastaut result was found twice, in separate trials, and the effect held across three different conditions.
One caveat: all four trials were funded by GW Pharmaceuticals, the drug’s developer, which was later acquired by Jazz Pharmaceuticals. Industry funding is standard for drug approval trials, and the results were peer reviewed and accepted by the FDA, but it is worth knowing.
The Dravet Syndrome Trial
Dravet syndrome is a genetic epilepsy that starts in infancy, with frequent, prolonged convulsive seizures and a high risk of early death. In the 2017 New England Journal of Medicine trial, 120 patients received CBD oral solution at 20 mg/kg/day or placebo for 14 weeks.
The median number of convulsive seizures per month fell by roughly half on CBD. Caregivers rated the patient’s overall condition as improved in 62% of the CBD group versus 34% on placebo. The share of patients whose seizures were cut by at least half was 43% versus 27%, a difference that did not reach statistical significance on its own. Five percent of CBD patients became free of convulsive seizures during the trial.
The Lennox-Gastaut Syndrome Trials
Lennox-Gastaut syndrome causes several seizure types, including “drop” seizures in which a person suddenly loses muscle tone and falls, often causing injuries. Both trials measured drop seizures.
GWPCARE3 compared two doses. Patients started with a median of 85 drop seizures a month. The 10 mg/kg dose worked almost as well as 20 mg/kg with fewer side effects, a finding that shaped the label: treatment usually starts low and goes to 10 mg/kg/day, with 20 mg/kg/day as the maximum. GWPCARE4 confirmed the effect at 20 mg/kg/day in 171 patients in the US, the Netherlands and Poland.
The Tuberous Sclerosis Complex Trial
Tuberous sclerosis complex (TSC) is a genetic disorder that causes benign growths in the brain and other organs, and mostly focal seizures. This made GWPCARE6 an important test of whether CBD worked beyond the two childhood syndromes. It did. At 25 mg/kg/day, seizures fell 48.6% versus 26.5% on placebo. The 50 mg/kg/day dose was no more effective and had more side effects, so 25 mg/kg/day became the recommended maintenance dose.
From Trials to FDA Approval
The FDA approved Epidiolex in June 2018 for seizures linked to Lennox-Gastaut and Dravet syndromes in patients 2 and older. It was the first FDA-approved drug containing a purified substance from the cannabis plant. In July 2020 the agency added tuberous sclerosis complex and lowered the age limit to 1 year. The current prescribing information reflects all three uses and states that Epidiolex is not a controlled substance.
Side Effects Seen in the Trials
Adverse events were more common on CBD than on placebo in every trial, though most were mild or moderate.
- Drowsiness, reduced appetite and diarrhea were the most frequent, and rose with dose. In the TSC trial, diarrhea affected 56% of patients at 50 mg/kg versus 25% on placebo.
- Liver enzyme elevations: 9% of CBD patients in GWPCARE3 and about 19% in GWPCARE6 had raised liver enzymes. The label says most cases occurred in patients also taking valproate, and the risk was also higher with clobazam.
- Stopping treatment: 14% of the CBD group in GWPCARE4 withdrew because of side effects, versus 1% on placebo.
Because of the liver risk, the label requires blood tests before starting and during treatment. It also carries warnings about sedation, suicidal thoughts, hypersensitivity and the danger of stopping seizure drugs abruptly.
What the Trials Do Not Prove
- They do not cover common epilepsies. Focal epilepsy in adults, the most common type, was not studied in these trials.
- They do not apply to marijuana or retail CBD. The drug is a purified, standardized product at weight-based doses (10 to 25 mg/kg/day) far above typical consumer use. Retail CBD is frequently mislabeled, as a 2017 JAMA analysis showed.
- They say nothing about THC. No THC was tested.
- They are not a cure. Seizure-free rates were low; CBD reduces seizures in some patients.
- They were short. The controlled phases ran 14 to 16 weeks; longer-term data come from open-label extensions without placebo groups.
Talk to Your Doctor
If your family is dealing with one of these syndromes, ask the neurologist whether Epidiolex fits, how dosing would be adjusted around current medicines such as valproate or clobazam, and how often liver tests are needed. Mention any CBD or cannabis already in use. Do not switch from a prescription to a dispensary product expecting the same results; the trials do not support that.
Epidiolex Research FAQ
How effective is Epidiolex for seizures?
In phase 3 trials, median seizure frequency fell about 37% to 49% on CBD versus about 15% to 27% on placebo, depending on the condition and dose. Results vary widely from person to person.
How long does Epidiolex take to work?
The trials measured results over 14 to 16 weeks, with doses increased gradually over the first weeks. Neurologists usually judge the response over several months.
Was the study that showed cannabis treats epilepsy reliable?
Older reports, including case studies of one or two patients, were not. The 2017 to 2021 Epidiolex trials were large, randomized and placebo-controlled, and they are the reason the FDA approved the drug.
Does Epidiolex contain THC?
It is purified CBD and is not a controlled substance under its FDA label. It does not cause a high.
Is Epidiolex approved for adults?
Yes. It is approved for patients 1 year and older with Lennox-Gastaut syndrome, Dravet syndrome or tuberous sclerosis complex, and adults were included in the Lennox-Gastaut and TSC trials.
For more studies on cannabis and the nervous system, browse our cannabis research section or read about cannabis and multiple sclerosis. For how CBD and THC are regulated, see our federal law guide.
For adults 21+ in legal states. General information, not medical advice. Talk to a doctor before using cannabis for a health condition.
