EN|UA
Trick or Treatment

Evidence Files

CBD Oil: Proven for Rare Epilepsy, at 700 Times the Dose EU Regulators Call Safe in Food

In the trials that made CBD a licensed medicine, patients with severe childhood-onset epilepsy took 20 mg per kilogram of body weight every day. In February 2026, the EU's food safety authority set a provisional safe intake for CBD food supplements of 0.0275 mg per kilogram, about 2 mg a day for an adult. The oils and gummies sold for calm, sleep and pain are backed by neither: their evidence is far thinner than the epilepsy data they borrow their reputation from.

Verdict

TRICK (weak evidence)

We searched PubMed and Cochrane Library for cannabidiol. There is plenty of research: 8,581 records with cannabidiol in the title or abstract, 347 of them tagged as randomised trials. For drug-resistant seizures in Dravet and Lennox-Gastaut syndromes the evidence is strong, and it produced a prescription medicine. For what consumer CBD is bought for, it is not: in a 2025 meta-analysis, CBD-only products did not significantly improve sleep, and the 2026 Cochrane review found no clear evidence that CBD-dominant medicines relieve neuropathic pain. The anxiety evidence rests on small studies at doses hundreds of times higher than a typical food-supplement intake.

Why Europe loves it

CBD arrived with the perfect pitch: the calming part of cannabis, minus the high. Across Europe it is sold as drops, gummies and capsules, with wording that implies calm, sleep or less pain without quite saying so.

The scientific credibility is real, but it belongs to a different product. Epidyolex, a purified CBD oral solution, received an EU marketing authorisation on 19 September 2019. It is prescription-only, approved for children from age 2 and adults with Lennox-Gastaut or Dravet syndrome (alongside clobazam) and, according to the EMA, also used for tuberous sclerosis complex.

Consumer CBD lives in a regulatory grey zone. In the EU, CBD in food counts as a novel food that needs pre-market authorisation. In 2022 the European Food Safety Authority put 19 pending applications on hold, concluding that the safety of CBD as a novel food could not be established because of data gaps on the liver, the gastrointestinal tract, the endocrine and nervous systems, and psychological function. In February 2026 it set a provisional safe level of about 2 mg a day for a 70 kg adult, and only for CBD of at least 98% purity. For people under 25, pregnant or breastfeeding women, and anyone taking medication, EFSA says safety cannot be established at all.

What the science actually says

The evidence for CBD splits cleanly in two: rigorous trials in rare epilepsies, at very high doses, and a much thinner literature for everything else.

  • Epilepsy: the real result. In a double-blind trial in the New England Journal of Medicine, 120 children and young adults with Dravet syndrome took 20 mg/kg of CBD a day or placebo on top of their usual drugs. Median monthly convulsive seizures fell from 12.4 to 5.9 with CBD, against 14.9 to 14.1 with placebo. In Lennox-Gastaut syndrome, a trial of 171 patients at sites in the USA, the Netherlands and Poland found drop seizures fell by a median 43.9% on CBD versus 21.8% on placebo. Both trials were funded by the manufacturer, GW Pharmaceuticals. Side effects were common: 14% of Lennox-Gastaut patients on CBD withdrew because of them, against 1% on placebo
  • Anxiety: small studies, big doses. The most cited study gave a single 600 mg dose to 12 people with social anxiety before a simulated public speech; they felt less anxious than 12 who got placebo. That dose is 300 times EFSA's provisional daily level. A 2024 meta-analysis pooled eight studies with 316 participants in total and found a significant effect, but its own authors urge caution because the clinical sample is so small
  • Expectation does part of the work. In a randomised crossover study, 43 healthy adults took CBD-free hempseed oil before a stress test, once told it contained CBD and once told it did not. Believing they had taken CBD changed sedation and heart-rate variability. Those who already believed CBD reduces anxiety reported feeling less anxious, from an oil that contained none
  • Sleep: no effect for CBD alone. A 2025 meta-analysis of six randomised trials with 1,077 participants found that cannabinoids as a group improved self-reported sleep quality. But the effect came from non-CBD cannabinoids. CBD-only products showed no significant effect (SMD 0.13, 95% CI −0.38 to 0.65)
  • Pain: the newest Cochrane review is negative. The 2026 update on chronic neuropathic pain pooled 21 trials with 2,187 people. For CBD-dominant medicines, five studies with 208 participants showed no clear evidence of at least 50% pain relief, on very low-certainty evidence. It covers neuropathic pain only
  • The liver signal shows up in healthy people too. In a phase I study run by the manufacturer's own researchers, 16 healthy adults took 1,500 mg of CBD a day for about three and a half weeks. Seven had liver enzymes above normal, and in five the rise met international criteria for drug-induced liver injury. That is a high dose, but it is the reason EFSA lists the liver first among its unanswered questions

Marketing claim vs. evidence

What the bottle implies

"Calm." "Restful sleep." "Recovery." Natural, non-psychoactive, from the same compound as an approved epilepsy medicine, and the label tells you exactly how many milligrams you are getting.

What the research shows

The epilepsy evidence is for a prescription drug at 20 mg/kg. CBD alone did not improve sleep in pooled trials. And the milligrams are unreliable: in a decade of testing in Europe, 49% of oils did not contain the CBD declared, and in the UK only 8% of products were within 10% of the label.

Frequently asked questions

+

Does CBD oil help with anxiety?

Possibly, but the evidence is weak: small studies, often single 600 mg doses before a lab stress test, and a 2024 meta-analysis with just 316 participants. People who believed in CBD also felt calmer after CBD-free oil.

+

Does CBD help you sleep?

On current evidence, CBD on its own does not. A 2025 meta-analysis found that cannabinoids improved sleep quality as a group, but the benefit came from non-CBD cannabinoids; CBD-only products showed no significant effect.

+

Does CBD oil work for pain?

For neuropathic pain, the 2026 Cochrane review found no clear evidence that CBD-dominant medicines give meaningful relief, based on five small trials. Products combining THC and CBD did slightly better, but the authors judged those effects not clinically relevant.

+

How much CBD is safe per day?

EFSA's provisional safe level for food supplements is 0.0275 mg per kilogram of body weight, about 2 mg a day for a 70 kg adult, for CBD of at least 98% purity. It does not apply to people under 25, during pregnancy or breastfeeding, or to anyone taking medication: for them, EFSA says safety cannot be established.

+

Can CBD oil damage your liver?

At high doses it can raise liver enzymes. In a study of healthy adults taking 1,500 mg a day, five of 16 developed elevations meeting criteria for drug-induced liver injury. Abnormal liver tests were also more common on CBD in the Dravet trial. Lower doses have been studied far less, which is exactly the gap EFSA points to.

+

Can CBD oil contain THC?

Yes. In a Czech analysis of 186 oils sold in Europe over the past decade, taking the recommended dose would exceed EFSA's acute reference dose for THC in 39% of cases. Some products also contained less familiar psychoactive compounds such as Δ8-THC and HHC.

+

Is CBD oil the same as Epidyolex?

No. Epidyolex is a prescription medicine dosed by body weight, tested at 20 mg/kg a day. Consumer CBD oil is a food supplement of variable content, and the epilepsy evidence does not transfer to it.

Our Conclusion

CBD is not snake oil. It is a genuine anticonvulsant, tested in rigorous, placebo-controlled trials and licensed in the EU for two severe childhood epilepsies. But that medicine is a purified, prescription-only drug given at around 20 mg per kilogram a day, with real side effects, including abnormal liver tests. None of that evidence transfers to a 10 ml bottle bought for stress. For what consumer CBD is actually sold for, the picture is weak: small anxiety studies at doses far above anything EFSA considers safe in food, no effect of CBD alone on sleep in pooled trials, and no clear benefit for neuropathic pain in the latest Cochrane review. Add labels that are wrong about half the time and occasional THC, and the honest question before buying is not "does CBD work?" but "does this product, at this dose, have any evidence for what I want it to do, and do I know what is actually in it?" For most bottles on the shelf, the answer to both is no.

Sources

Every claim in the text is linked to a primary source. Links go to the publisher's site, to the European Medicines Agency or EFSA, or to PubMed, the US National Library of Medicine database.

  1. Devinsky O, Cross JH, Laux L, et al. Trial of Cannabidiol for Drug-Resistant Seizures in the Dravet Syndrome. The New England Journal of Medicine, 2017;376(21):2011–2020.
  2. Thiele EA, Marsh ED, French JA, et al. Cannabidiol in patients with seizures associated with Lennox-Gastaut syndrome (GWPCARE4): a randomised, double-blind, placebo-controlled phase 3 trial. The Lancet, 2018;391(10125):1085–1096.
  3. Ateş G, Welsch P, Klose P, et al. Cannabis-based medicines for chronic neuropathic pain in adults. Cochrane Database of Systematic Reviews, 2026;1(1):CD012182.
  4. da Silva GHS, Barbosa EC, de Lima FR, et al. Effectiveness of cannabinoids on subjective sleep quality in people with and without insomnia or poor sleep: A systematic review and meta-analysis of randomised studies. Sleep Medicine Reviews, 2025;84:102156.
  5. Bergamaschi MM, Queiroz RH, Chagas MH, et al. Cannabidiol reduces the anxiety induced by simulated public speaking in treatment-naïve social phobia patients. Neuropsychopharmacology, 2011;36(6):1219–26.
  6. Han K, Wang JY, Wang PY, Peng YC. Therapeutic potential of cannabidiol (CBD) in anxiety disorders: A systematic review and meta-analysis. Psychiatry Research, 2024;339:116049.
  7. Spinella TC, Stewart SH, Naugler J, et al. Evaluating cannabidiol (CBD) expectancy effects on acute stress and anxiety in healthy adults: a randomized crossover study. Psychopharmacology, 2021;238(7):1965–1977.
  8. Watkins PB, Church RJ, Li J, Knappertz V. Cannabidiol and Abnormal Liver Chemistries in Healthy Adults: Results of a Phase I Clinical Trial. Clinical Pharmacology & Therapeutics, 2021;109(5):1224–1231.
  9. Binova Z, Maly M, Drabova L, et al. Poorly labeled CBD and related cannabinoid oils marketed in Europe: a hidden risk for consumers? Insights from a long-term study. Analytical and Bioanalytical Chemistry, 2026.
  10. Johnson DA, Hogan M, Marriot R, et al. A comparison of advertised versus actual cannabidiol (CBD) content of oils, aqueous tinctures, e-liquids and drinks purchased in the UK. Journal of Cannabis Research, 2023;5(1):28.
  11. EFSA Panel on Nutrition, Novel Foods and Food Allergens. Statement on safety of cannabidiol as a novel food: data gaps and uncertainties. EFSA Journal, 2022;20(6):7322.
  12. European Food Safety Authority. Provisional safe level for cannabidiol as a novel food. 9 February 2026.
  13. European Medicines Agency. Epidyolex (cannabidiol): EPAR.
Share:

This article is for informational purposes only and does not constitute medical advice. Trick or Treatment analyses the presence of clinical studies in open scientific databases — PubMed and Cochrane Library. The absence of studies in these databases does not automatically mean a drug is ineffective, but it does mean its effectiveness has not been confirmed by evidence-based medicine standards. Any treatment decisions should be made together with your doctor.