Cannabidiol (CBD) is a cannabinoid that occurs naturally in the Cannabis sativa plant. Unlike THC (delta-9-tetrahydrocannabinol), it does not produce a “high”: in a human abuse-potential study, single doses up to 4,500 mg scored in the placebo range for drug liking (Epidiolex label). CBD is sold in two very different forms: as Epidiolex, a prescription drug for three rare seizure disorders, and as unregulated oils, gummies, vapes and creams marketed for anxiety, sleep, pain and recovery.
The most important fact is the gap between the two. Epidiolex is backed by randomized trials at doses of 10–25 mg/kg/day. The consumer products are not. Placebo-controlled trials of CBD for pain, sleep and exercise recovery have mostly been negative. Lab tests of retail products have found wrong CBD amounts and unlabeled THC. The FDA’s position is that CBD cannot legally be sold as a dietary supplement or added to food.
This page summarizes what the Epidiolex label, clinical trials and regulators report. It is for education only and is not medical advice.
| Property | Detail |
|---|---|
| Class | Plant cannabinoid; anticonvulsant (Epidiolex) |
| Approved product | Epidiolex (cannabidiol) oral solution, 100 mg/mL; initial US approval 2018 |
| Labeled indications | Seizures of Lennox-Gastaut syndrome, Dravet syndrome or tuberous sclerosis complex, age 1 year and older |
| Labeled dose | 5 mg/kg/day to start; 10–20 mg/kg/day maintenance (LGS, Dravet); 25 mg/kg/day (TSC) |
| Half-life (label) | 56–61 h after twice-daily dosing for 7 days |
| Mechanism | Anticonvulsant mechanism unknown; does not appear to act through cannabinoid receptors (label) |
| Key label warning | Dose-related liver enzyme rises, much more common with valproate |
| US status | Epidiolex is not a controlled substance. FDA says CBD is excluded from the dietary supplement definition |
| WADA status | Not prohibited: CBD is the only named exception in S8 Cannabinoids; THC is prohibited in-competition |
What it is
CBD’s chemical formula is C₂₁H₃₀O₂, molecular weight 314.46. It is a white to pale yellow crystalline solid that does not dissolve in water (Epidiolex label). Because it dissolves only in fats and solvents, both Epidiolex and most retail products are oil-based. Epidiolex uses sesame seed oil with 7.9% alcohol.
CBD is often described as acting on the “endocannabinoid system,” but the label is more cautious. It states that the precise mechanism behind its anticonvulsant effect in humans is unknown, and that CBD does not appear to work through cannabinoid receptors. That is one reason it lacks THC’s intoxicating effects.
The two product worlds differ in ways that matter:
| Feature | Epidiolex | Retail “CBD” products |
|---|---|---|
| Regulation | FDA-approved drug, manufactured to drug standards | Not FDA-approved; FDA says they cannot be sold as supplements or foods |
| Content | 100 mg/mL purified CBD | Often differs from the label (see Quality and contamination) |
| THC | Purified product; no THC-like drug liking in abuse study | THC found in about 1 in 5 products in one lab survey |
| Evidence | Four randomized trials in severe epilepsy | Small trials, mostly negative or inconsistent |
Pharmacokinetics
All figures come from the Epidiolex label (DailyMed):
| Parameter | Value |
|---|---|
| Time to peak (steady state) | 2.5–5 h |
| Dose proportionality | Exposure rises less than proportionally from 5 to 25 mg/kg/day |
| High-fat meal | Peak level about 5×, total exposure (AUC) about 4× higher than fasting |
| Low-fat meal | Peak about 4×, AUC about 3× higher |
| Alcohol | Peak 93% higher, AUC 63% higher |
| Protein binding | Over 94% |
| Half-life | 56–61 h (twice-daily dosing for 7 days, healthy subjects) |
| Metabolism | Liver and gut: CYP2C19, CYP3A4, UGT1A7, UGT1A9, UGT2B7 |
| Metabolites | 7-OH-CBD (active; AUC 38% lower than CBD); 7-COOH-CBD (AUC about 40× higher than CBD) |
| Excretion | Mainly in feces; minor renal clearance |
| Moderate or severe liver impairment | AUC about 2.5–5.2× higher; label lowers doses |
Two points stand out. First, food changes exposure several-fold, which is why the label asks for consistent dosing relative to meals. Second, the long half-life means CBD builds up over several days of regular use.
Approved medical use
Epidiolex is the only FDA-approved drug containing CBD (FDA). It is indicated for seizures associated with Lennox-Gastaut syndrome (LGS), Dravet syndrome or tuberous sclerosis complex (TSC) in patients 1 year and older. The label requires liver tests (ALT, AST and bilirubin) before starting.
| Indication | Start | Maintenance | Maximum |
|---|---|---|---|
| LGS or Dravet syndrome | 2.5 mg/kg twice daily (5 mg/kg/day) | 5 mg/kg twice daily (10 mg/kg/day) after 1 week | 10 mg/kg twice daily (20 mg/kg/day) |
| TSC | 2.5 mg/kg twice daily (5 mg/kg/day) | Raise weekly to 12.5 mg/kg twice daily (25 mg/kg/day) | 25 mg/kg/day is the recommended dose |
The label notes that 20 mg/kg/day gave somewhat greater seizure reductions than 10 mg/kg/day, but with more adverse reactions. Doses are lower in moderate or severe liver impairment. Like other antiseizure drugs, Epidiolex should be tapered rather than stopped abruptly.
The pivotal trials
All four trials were randomized, double-blind and placebo-controlled, with CBD added to existing antiseizure drugs. Results are median percent changes in seizure frequency per 28 days, from the label:
| Trial | n | Placebo | CBD 10 mg/kg/day | CBD 20 mg/kg/day | CBD 25 mg/kg/day |
|---|---|---|---|---|---|
| LGS, Study 1 (drop seizures) | 171 | −22% | — | −44% | — |
| LGS, Study 2 (drop seizures) | 225 | −17% | −37% | −42% | — |
| Dravet, Study 3 (convulsive seizures) | 120 | −13% | — | −39% | — |
| TSC, Study 4 (TSC seizures) | 224 | −20% | — | — | −43% |
In the Dravet trial, 43% of patients on CBD had at least a 50% drop in convulsive seizures, versus 27% on placebo (odds ratio 2.00, 95% CI 0.93–4.30, not significant) (Devinsky et al., NEJM 2017). Few patients became seizure-free: 5% on CBD vs 0% on placebo. CBD is an add-on that reduces seizures in these syndromes; it does not cure them.
What the research shows for other uses
Outside epilepsy, the evidence is thin. Most trials are small, short and use different products and doses, so results are hard to compare.
Anxiety
The best-known studies used a laboratory stress test, the simulated public speaking test, with a single dose given before the speech:
| Study | Participants | CBD dose | Result |
|---|---|---|---|
| Bergamaschi et al., 2011 | 24 untreated people with social anxiety disorder | 600 mg once vs placebo | Less anxiety, cognitive impairment and discomfort during the speech |
| Linares et al., 2019 | 57 healthy men | 150, 300 or 600 mg once vs placebo | Only 300 mg reduced anxiety; 150 and 600 mg did not differ from placebo |
The Linares result suggests an “inverted U” dose response, where more is not better. These are single-dose lab studies, not treatment trials.
A 2025 meta-analysis pooled 21 placebo-controlled trials of cannabinoids in anxiety-related disorders and Tourette syndrome (Raminelli et al., Cannabis Cannabinoid Res 2025). For pure or CBD-enriched products the pooled effect was not significant (SMD −0.40, 95% CI −0.84 to 0.03). A subgroup limited to pure CBD showed a moderate effect (SMD −0.61, 95% CI −1.15 to −0.07). The authors rated the evidence as low quality and limited by small samples and mixed designs.
Sleep
In a randomized pilot trial, 30 adults with moderate-to-severe insomnia took 150 mg CBD or placebo under the tongue nightly for 2 weeks (Narayan et al., J Clin Sleep Med 2024). Results:
- No difference in insomnia severity, self-reported time to fall asleep, sleep efficiency or time awake after sleep onset.
- Higher well-being scores on CBD, and better wrist-measured sleep efficiency after 2 weeks.
The authors concluded CBD was similar to placebo on most sleep outcomes. The Epidiolex label reports somnolence as a common side effect at 10–25 mg/kg/day, but that is sedation in epilepsy patients, not evidence of better sleep.
Pain
Three placebo-controlled trials, at doses from 20 mg to 600 mg per day, found no pain benefit:
| Trial | Condition, n | CBD dose and duration | Primary result |
|---|---|---|---|
| Vela et al., Pain 2022 | Hand osteoarthritis or psoriatic arthritis, 136 | Synthetic CBD 20–30 mg/day, 12 weeks | Difference 0.23 mm on a 0–100 mm pain scale (95% CI −9.41 to 9.90); no effect on sleep, anxiety or depression scores |
| Pramhas et al., Lancet Reg Health Eur 2023 | Knee osteoarthritis on paracetamol 3 g/day, 86 | 600 mg/day, 8 weeks | WOMAC pain fell 2.5 points on CBD vs 2.4 on placebo (p = 0.80) |
| Bebee et al., Med J Aust 2021 | Acute low back pain in the emergency department, 100 | 400 mg once | Pain at 2 h 6.2 vs 5.8 of 10; no difference in rescue opioid use |
In the knee trial, more adverse events occurred with CBD (135 vs 105 events), and rises in liver enzymes were more common (15 vs 5 patients, p = 0.02).
Exercise performance and recovery
Athletes often use CBD for recovery, but controlled data show little effect:
- Recovery after hard training. In a double-blind crossover trial, 17 well-trained athletes took 60 mg CBD (oil or a water-soluble form) or placebo after each session of a 6-day high-intensity program (Isenmann et al., JISSN 2024). Neither product changed performance or inflammatory markers. The only signal was lower myoglobin with the oil in one subgroup, and the authors found no clear effects overall.
- Endurance. In a randomized crossover trial, 25 trained runners took placebo, 50 mg or 300 mg CBD 1.5 h before a 60-minute run and a run to exhaustion (Sahinovic et al., Sports Med Open 2025). Neither dose changed time to exhaustion, perceived exertion, pain, mood or muscle-damage markers. The authors concluded CBD is unlikely to help or hurt performance at these doses.
Side effects and risks
Liver injury
The Epidiolex label’s main warning is dose-related elevation of liver enzymes (DailyMed). In controlled trials, ALT rose above 3 times the upper limit of normal (ULN) in 13% of patients on 10–20 mg/kg/day and 12% on 25 mg/kg/day, versus 1% on placebo. Fewer than 1% exceeded 20 times the ULN. Risk depended strongly on other drugs:
| LGS or Dravet patients on CBD | ALT over 3× ULN |
|---|---|
| Valproate and clobazam | 30% |
| Valproate without clobazam | 21% |
| Clobazam without valproate | 4% |
| Neither drug | 3% |
| By dose: 10 mg/kg/day | 1% |
| By dose: 20 mg/kg/day | 17% |
Most rises appeared in the first 2 months, but some occurred up to 18 months in, particularly with valproate. The label calls for ALT, AST and bilirubin at baseline and at 1, 3 and 6 months. It says to stop CBD if transaminases exceed 3× ULN with bilirubin over 2× ULN, or stay above 5× ULN.
Liver effects are not limited to epilepsy patients on other drugs:
- FDA trial at a consumer-range dose. FDA scientists randomized 201 healthy adults to CBD 5 mg/kg/day (about 400 mg/day) or placebo for 28 days (Florian et al., JAMA Intern Med 2025). ALT or AST rose above 3× ULN in 8 of 151 on CBD (5.6%) and 0 of 50 on placebo. Seven met criteria for possible drug-induced liver injury, all with eosinophilia (raised levels of a type of white blood cell), and enzymes returned to normal within 1–2 weeks of stopping.
- High-dose study. In 16 healthy adults taking 1,500 mg/day for about 3.5 weeks, 7 (44%) had ALT above the ULN and 5 (31%) above 5× ULN (Watkins et al., Clin Pharmacol Ther 2021).
Other labeled adverse effects
From the Epidiolex label, in controlled trials in LGS and Dravet syndrome unless noted:
| Effect | CBD | Placebo |
|---|---|---|
| Somnolence and sedation | 32% (46% with clobazam, 16% without) | 11% |
| Weight loss of 5% or more | 16% (31% in TSC trial) | 8% |
| New anemia | 30% (38% in TSC trial) | 13% (15% in TSC trial) |
| Pneumonia with clobazam, 20 mg/kg/day | 10% | 1% |
Other labeled effects include decreased appetite, diarrhea, rash and infections. The label also notes:
- Serum creatinine: a rise of about 10% within 2 weeks, which a dedicated study showed is not due to lower kidney filtration.
- Hypersensitivity: itching, redness and angioedema (swelling under the skin) occurred in trials. Epidiolex contains sesame seed oil and is contraindicated in people allergic to its ingredients.
- Suicidal thoughts: a class warning for all antiseizure drugs.
- QTc: a concentration-dependent prolongation of the heart’s QT interval, not more than 10 ms at recommended monotherapy doses taken with a high-fat meal.
Hormones, pregnancy and dependence
The FDA has cited “possible harm to the male reproductive system” as a concern (FDA, 2023). The label’s evidence for this comes from animals: in rats, CBD given through pregnancy and lactation caused delayed sexual maturation and small testes in male offspring. In the 28-day FDA trial in healthy adults, CBD did not change total testosterone or inhibin B in men (Florian et al., 2025). Longer human data are lacking.
Pregnant animals given CBD showed developmental toxicity, and there are no adequate human pregnancy data. In humans, 1,500 mg/day for 28 days caused no withdrawal symptoms after stopping, suggesting CBD likely does not cause physical dependence (label).
Interactions
CBD changes the levels of many drugs, and others change CBD. From the Epidiolex label:
| Drug or group | Effect | Label advice |
|---|---|---|
| Clobazam | Active metabolite N-desmethylclobazam about 3× higher | Consider lowering clobazam |
| Valproate | More liver enzyme rises | Reduce or stop CBD and/or valproate if enzymes rise |
| CYP2C19 substrates | CBD is a moderate CYP2C19 inhibitor; prodrugs such as clopidogrel may lose active metabolite | Adjust dose as appropriate |
| CYP1A2 substrates (e.g. theophylline, tizanidine) | Caffeine AUC rose 95% with CBD 750 mg twice daily | Consider dose reduction |
| Everolimus and other oral P-gp substrates (sirolimus, tacrolimus, digoxin) | Everolimus exposure about 2.5× higher | Drug-level monitoring, lower doses |
| UGT1A9 substrates (e.g. mycophenolate) | Mycophenolic acid AUC 35% higher | Consider dose reduction |
| Stiripentol | Stiripentol exposure higher | Monitor |
| Strong CYP3A4/CYP2C19 inducers (e.g. rifampin) | CBD exposure 32% lower | Consider up to 2× CBD dose |
| Alcohol and other CNS depressants | More sedation; alcohol also raises CBD levels | Caution |
These interactions come from studies of the prescription product. The same molecule is in retail products, but at unknown and variable doses.
Quality and contamination
The best-known label audit tested 84 CBD extracts from 31 companies bought online in 2016 (Bonn-Miller et al., JAMA 2017). Accurate labeling was defined as within ±10% of the stated CBD content:
| Finding | Products |
|---|---|
| Accurately labeled | 26 of 84 (31%) |
| More CBD than labeled | 36 of 84 (43%) |
| Less CBD than labeled | 22 of 84 (26%) |
| THC detected (up to 6.43 mg/mL) | 18 of 84 (21%) |
Vape liquids were most often mislabeled (88%), and oils most often accurate (45%). The authors noted the THC found could be enough to cause intoxication, especially in children.
Small amounts of THC can also cause positive drug tests. In a controlled study, 60 adults took 100 mg CBD in oil with 0–3.7 mg THC, twice daily for 14 days (Wolinsky et al., J Anal Toxicol 2026). At least one participant in every THC-containing group had a positive urine test during the 14 days, even at 0.5 mg. One sample in the CBD-only group was also positive. The THC groups showed no differences from CBD alone in drug effects or impairment measures. The authors note that standard drug tests cannot tell legal hemp products from cannabis use.
Legal status and use in sport
FDA
The FDA has concluded that CBD products are excluded from the legal definition of a dietary supplement, because CBD is the active ingredient in an approved drug (21 U.S.C. 321(ff)(3)(B)). For the same reason it says adding CBD to food sold across state lines is prohibited (FDA Q&A). The FDA has issued warning letters to companies claiming CBD treats diseases such as cancer.
In January 2023, the FDA stated that existing food and supplement rules are not appropriate for CBD. It said it had not found adequate evidence on how much CBD can be consumed, and for how long, before causing harm. It denied three petitions to allow CBD supplements and asked Congress for a new regulatory pathway (FDA statement, 2023).
DEA and the Farm Bill
- 2018: DEA placed FDA-approved CBD drugs with no more than 0.1% THC in Schedule V in September 2018 (DEA, 2026).
- 2018 Farm Bill: signed December 20, 2018. It removed “hemp” from the Controlled Substances Act: cannabis and its derivatives, including cannabinoids, with no more than 0.3% delta-9 THC on a dry-weight basis (FDA Q&A).
- 2020: DEA removed the Schedule V listing for approved CBD drugs, since they were no longer controlled under the Farm Bill. It also stated that CBD mixtures with more than 0.3% delta-9 THC remain Schedule I (DEA interim final rule, 2020). The current Epidiolex label states it is not a controlled substance.
- 2025 law, effective November 12, 2026: Section 781 of Public Law 119-37 rewrites the federal hemp definition (govinfo). The THC limit becomes 0.3% total THC, including THCA. Final consumer products with more than 0.4 mg of combined total THC (and THC-like cannabinoids) per container are excluded from “hemp,” as are products with synthesized cannabinoids. FDA-approved drugs are excluded from the new product definitions. A 2026 DEA rule confirms the November 12, 2026 effective date (DEA, 2026).
The same April 2026 DEA rule moved FDA-approved drugs containing marijuana, and state-licensed medical marijuana, to Schedule III. It states that it does not change the status of hemp. State laws on CBD vary and are not covered here.
WADA
The WADA 2026 Prohibited List prohibits all natural and synthetic cannabinoids in-competition under S8, with one named exception: cannabidiol. THC is listed as a “Substance of Abuse.” Other plant cannabinoids are not exempt.
For athletes, the practical risk is contamination. A product sold as CBD may contain THC that is not on the label (21% of products in the 2017 JAMA survey), and as little as 0.5 mg of THC twice daily produced positive urine tests in the 2026 dosing study (using workplace drug-testing cutoffs).
Key takeaways
- One approved drug, one narrow use. Epidiolex is FDA-approved for seizures in Lennox-Gastaut syndrome, Dravet syndrome and tuberous sclerosis complex, at 10–25 mg/kg/day. In trials it cut seizure frequency by a median of 37–44%, versus 13–22% on placebo.
- Weak evidence for popular uses. Placebo-controlled trials found no benefit for arthritis pain, acute back pain, insomnia or exercise recovery. Anxiety data come mostly from small single-dose lab studies.
- The liver is the main safety issue. ALT rose above 3× normal in 12–13% of Epidiolex patients (up to 30% with valproate and clobazam), and in 5.6% of healthy adults at about 400 mg/day in an FDA trial.
- Many drug interactions. CBD raises levels of clobazam’s active metabolite, caffeine, everolimus and other drugs, and interacts with valproate.
- Retail products are unreliable. Only 31% of tested online extracts matched their label, and 21% contained THC.
- Allowed in sport, but contamination is a risk. WADA exempts CBD from S8, but THC is prohibited in-competition, and trace THC in hemp products can cause positive urine tests.
Sources
- Jazz Pharmaceuticals. Epidiolex (cannabidiol) oral solution prescribing information. Revised 2026. DailyMed.
- Devinsky O, Cross JH, Laux L, et al. Trial of cannabidiol for drug-resistant seizures in the Dravet syndrome. N Engl J Med. 2017;376(21):2011–2020. PubMed.
- 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):1219–1226. PubMed.
- Linares IM, Zuardi AW, Pereira LC, et al. Cannabidiol presents an inverted U-shaped dose-response curve in a simulated public speaking test. Braz J Psychiatry. 2019;41(1):9–14. PubMed.
- Raminelli AO, Simei JLQ, Guimarães FS, et al. Effects of different cannabinoid formulations on anxiety-related disorders, and Tourette syndrome: a systematic review and meta-analysis. Cannabis Cannabinoid Res. 2025;10(6):655–672. PubMed.
- Narayan AJ, Downey LA, Rose S, et al. Cannabidiol for moderate-severe insomnia: a randomized controlled pilot trial of 150 mg of nightly dosing. J Clin Sleep Med. 2024;20(5):753–763. PubMed.
- Vela J, Dreyer L, Petersen KK, et al. Cannabidiol treatment in hand osteoarthritis and psoriatic arthritis: a randomized, double-blind, placebo-controlled trial. Pain. 2022;163(6):1206–1214. PubMed.
- Pramhas S, Thalhammer T, Terner S, et al. Oral cannabidiol (CBD) as add-on to paracetamol for painful chronic osteoarthritis of the knee: a randomized, double-blind, placebo-controlled clinical trial. Lancet Reg Health Eur. 2023;35:100777. PubMed.
- Bebee B, Taylor DM, Bourke E, et al. The CANBACK trial: a randomised, controlled clinical trial of oral cannabidiol for people presenting to the emergency department with acute low back pain. Med J Aust. 2021;214(8):370–375. PubMed.
- Isenmann E, Veit S, Flenker U, et al. Influence of short-term chronic oral cannabidiol application on muscle recovery and performance after an intensive training protocol – a randomized double-blind crossover study. J Int Soc Sports Nutr. 2024;21(1):2337252. PubMed.
- Sahinovic A, Lau NS, Sabag A, et al. The acute effects of cannabidiol on physiological and subjective responses to endurance exercise: a dose-ranging randomised controlled crossover trial. Sports Med Open. 2025;11(1):96. PubMed.
- Florian J, Salcedo P, Burkhart K, et al. Cannabidiol and liver enzyme level elevations in healthy adults: a randomized clinical trial. JAMA Intern Med. 2025;185(9):1070–1078. PubMed.
- Watkins PB, Church RJ, Li J, Knappertz V. Cannabidiol and abnormal liver chemistries in healthy adults: results of a phase I clinical trial. Clin Pharmacol Ther. 2021;109(5):1224–1231. PubMed.
- Bonn-Miller MO, Loflin MJE, Thomas BF, et al. Labeling accuracy of cannabidiol extracts sold online. JAMA. 2017;318(17):1708–1709. PubMed.
- Wolinsky D, Zamarripa CA, Spindle TR, et al. The acute and chronic pharmacokinetics and pharmacodynamics of oral cannabidiol with and without low doses of delta-9-tetrahydrocannabinol. J Anal Toxicol. 2026;50(1):bkaf075. PubMed.
- US Food and Drug Administration. FDA regulation of cannabis and cannabis-derived products, including cannabidiol (CBD): questions and answers. Content current as of July 16, 2024. FDA.
- Woodcock J. FDA concludes that existing regulatory frameworks for foods and supplements are not appropriate for cannabidiol, will work with Congress on a new way forward. FDA statement, January 26, 2023. FDA.
- Drug Enforcement Administration. Implementation of the Agriculture Improvement Act of 2018. Interim final rule. Federal Register. August 21, 2020;85:51639 (FR Doc 2020-17356). Federal Register.
- Continuing Appropriations, Agriculture, Legislative Branch, Military Construction and Veterans Affairs, and Extensions Act, 2026. Public Law 119-37, section 781. November 12, 2025;139 Stat. 495. govinfo.
- Drug Enforcement Administration. Schedules of controlled substances: rescheduling of Food and Drug Administration approved products containing marijuana from schedule I to schedule III; corresponding change to permit requirements. Final rule. Federal Register. April 28, 2026;91:22714 (FR Doc 2026-08176). Federal Register.
- World Anti-Doping Agency. The 2026 Prohibited List, World Anti-Doping Code. Valid 1 January 2026. WADA 2026 Prohibited List.