Cannabis and Epilepsy: What Charlotte's Web and Epidiolex Actually Proved

A five-year-old girl in Colorado went from around three hundred seizures a week to two or three a month. Her parents said it was the cannabis oil. A CNN documentary in 2013 put the story in front of millions, sixteen states rewrote their laws, and an entire industry was born on it.

The girl's name was Charlotte Figi. Her case was never written up in the medical literature.

That is not a reason to dismiss the story. It is the reason the story needed testing, and five years later it got tested properly. This guide covers what those trials found, why the results are both real and smaller than the legend, and the drug interaction that complicates every number in them.

What was claimed, and what was documented

The Figi family's account is consistent and has never wavered. Charlotte had Dravet syndrome, a severe genetic epilepsy that begins in infancy and resists nearly every anticonvulsant. Her seizure count was catastrophic. After starting a high-CBD, low-THC oil grown by the Stanley brothers, it collapsed.

What did not exist was documentation. In 2014, Colorado's then chief medical officer put it flatly: we don't have any peer-reviewed, published literature to support it.

In the same reporting, a vice president of the American Epilepsy Society made the point that matters most for interpreting any single case. Seizure frequency fluctuates naturally, sometimes dramatically, which is exactly why anecdotal recoveries mislead. A child can improve for reasons nobody can identify, and the treatment started that month gets the credit.

Charlotte died in April 2020, aged 13, after a severe seizure and respiratory failure complicated by pneumonia. She had been treated as a likely COVID-19 case during a hospitalization, though her test on April 5 came back negative. Her mother said afterward that she was seizure-free forever.

The drug that came out of it

The strongest vindication of the underlying idea is not a strain. It is a pharmaceutical.

Epidiolex, a purified plant-derived cannabidiol, was approved by the FDA on June 25, 2018 for Lennox-Gastaut syndrome and Dravet syndrome. It has since been approved for tuberous sclerosis complex as well, in patients one year of age and older.

The pivotal trial numbers, from the label:

• Lennox-Gastaut, 171 patients: 44% median reduction in drop seizures at 20 mg/kg/day against 22% on placebo, p equal to 0.01
• Lennox-Gastaut, 225 patients: 37% at 10 mg/kg/day and 42% at 20 mg/kg/day, against 17% on placebo
• Dravet, 120 patients: 39% median reduction in convulsive seizures against 13% on placebo, p equal to 0.01
• Tuberous sclerosis, 224 patients: 43% against 20% on placebo

Those are real effects, replicated across four trials with placebo controls. They are also not the story people carry around. The placebo groups improved by 13% to 22%, which is precisely the natural fluctuation the AES warned about. The drug's advantage over placebo is roughly twenty percentage points, not the near-elimination of seizures in the anecdote.

The scheduling history closed the loop. Epidiolex went into Schedule V in September 2018, and in April 2020 the DEA removed it from the Controlled Substances Act entirely.

The complication nobody mentions

Here is the part that almost never appears in coverage, and it changes how you should read every figure above.

Around half the patients in the Lennox-Gastaut trials, and 65% in the Dravet trial, were also taking clobazam, a benzodiazepine. CBD raises clobazam levels substantially. The label carries a warning that concomitant use increases plasma concentrations of the active metabolite and advises considering a clobazam dose reduction.

So the obvious question is whether some of the effect was CBD boosting an existing drug instead of working on its own. A meta-analysis answered it with numbers. Among patients not taking clobazam, 29.1% on CBD achieved at least a 50% seizure reduction against 15.7% on placebo. Among patients on clobazam, it was 52.9% against 27.8%.

CBD works without clobazam. The effect is smaller. Both halves of that sentence are true and only one of them usually gets said.

That same analysis found CBD raises clobazam levels by around 60% and its active metabolite by up to 500%, which is also why the sedation people report is not always the CBD.

What it does to the liver

Elevated liver enzymes are the most common reason patients stop taking it. In the trials, ALT above three times the upper limit of normal appeared in 13% of Lennox-Gastaut and Dravet patients and 12% of tuberous sclerosis patients, against 1% on placebo.

The stratification is the useful part. Among patients on both valproate and clobazam, that rate was 30%. On valproate alone, 21%. On clobazam alone, 4%. On neither, 3%.

Most of the liver signal is an interaction with valproate, not CBD in isolation.

The unexamined day is a wasted opportunity. Reflect on what you did, what you learned, and how you can improve.

John Dewey

Shelf CBD is not the same product

This is where the gap between the trials and the bottle in the pharmacy aisle becomes a safety issue.

The Epilepsy Foundation is explicit that the trials used pharmaceutical-grade CBD and that there is significantly less information on artisanal marijuana, which may have less consistent doses and potentially harmful impurities, noting that products are not uniform from batch to batch.

The American Epilepsy Society goes further, warning that dispensary products may contain other phytocannabinoids including THC alongside pesticides and other impurities at unknown concentrations, and that anecdotal reports alone are not sufficient to support treatment decisions.

Testing bears that out. An analysis of 84 CBD products from 31 companies found only 30% contained CBD within 10% of the labeled amount, with 42% under-labeled and 26% over-labeled, and some containing meaningful THC.

For a child on a weight-based dose, a bottle that contains 40% less than the label says is not a minor inconvenience. It is the difference between a therapeutic dose and nothing.

What the evidence does not cover

THC. There is no trial evidence that THC reduces seizures, and the major epilepsy organizations do not present any. In their materials THC appears as the psychoactive component and as an impurity risk, not as a treatment.

Epilepsy types outside the three approved indications are also thinly evidenced. Open-label work published in 2025 followed 140 patients with treatment-resistant focal epilepsies and reported median seizure reductions of 54% to 77% across 144 weeks, which sounds spectacular until you note there was no placebo arm and 38% of participants discontinued. A Phase 3 trial in Japan missed its primary endpoint outright.

Adults with common focal epilepsies from stroke or head injury are barely represented in the trial literature at all.

The strain that does not exist

We want to be straight about something a seed company is not supposed to say.

You cannot buy a CBD-rich strain from us on the US market. There is no high-CBD line in our American catalog, and the reason is a market one: American demand went almost entirely to THC, and CBD-dominant genetics stopped selling.

That matters for this article because it tells you something about the whole category. Charlotte's Web was roughly 17% CBD against 0.5% THC. Producing a plant with that ratio takes deliberate multi-generation selection for a trait most buyers actively avoid, and when the buyers vanish, so does the breeding program.

What we can tell you is what a modern line's cannabinoid content honestly looks like. Wedding Cake Auto is the only strain in our US catalog with a published CBD figure, at 2.4% alongside 26% THC, running 70 to 75 days from seed at 80 to 100 cm. That 2.4% is a rounding error next to Charlotte's Web and we are not going to pretend otherwise. Afghan Hash Plant is the closest thing we sell to a pre-potency-race plant, a pure indica at 21% collected from the Mazari Sharif region of northern Afghanistan in the 1970s, finishing in 50 to 60 days.

Neither is a medicine and neither is close to what those trials used. Anyone treating epilepsy needs a pharmaceutical with a certificate of analysis and a neurologist, not a seed catalog.

The short version

Charlotte Figi's case was never documented in the literature, and the doctors who said so at the time were right to. The trials that followed found a real, replicated, moderate effect for purified CBD in three specific severe epilepsies, roughly twenty percentage points better than placebo.

Around half of trial participants were also on clobazam, and CBD's benefit is visibly larger in that group, though it persists without it. Liver enzyme elevations track mostly with valproate co-medication. Retail CBD is mislabeled more often than not.

There is no evidence for THC and seizures, and thin evidence outside the three approved indications. If someone in your family has epilepsy, the version that works is the one with an FDA label on it.

Barney's Farm has been developing premium cannabis genetics since the 1980s, with over 40 Cannabis Cup wins. Explore our full seed catalog and find strains bred for every climate and skill level.