Cannabidiol (CBD)
Summary
CBD is a cannabis compound that does not cause a high. Single 300–600 mg doses lowered anxiety in small public-speaking trials (grade C), but a 2-week insomnia pilot found no change in insomnia severity. At 1,500 mg/day, 5 of 16 healthy adults met liver-injury criteria, and CBD raises levels of drugs such as tacrolimus. The FDA says CBD cannot be sold as a dietary supplement.
Key facts
- A non-intoxicating cannabinoid; purified CBD is the active ingredient in the prescription drug Epidiolex[2][1]
- FDA: CBD products are excluded from the dietary supplement definition, and CBD cannot be added to food sold across state lines[1]
- Anxiety in public speaking tests (C); scan-related anxiety (mixed, D); insomnia (mixed, D)[8][4][5][6][7]
- Single 150–600 mg doses for anxiety; 150 mg nightly for 2 weeks for insomnia; prescription dosing of 5–25 mg/kg/day[4][5][7][3]
- At 1,500 mg/day, 7 of 16 healthy adults had ALT above normal and 5 met drug-induced liver injury criteria[9]
- Raised tacrolimus peak levels 4.2-fold and roughly doubled caffeine exposure in pharmacokinetic studies[11][12]
- GW Research (Epidiolex maker) sponsored or employed authors of the liver and caffeine studies; Cannvalate funded the insomnia pilot; the dose-response trial authors hold a CBD-compound patent[9][12][7][5]
- 6 graded human outcomes on this page
- October 10, 2026
How it works
ProposedCBD acts on several brain signalling systems rather than mainly on the cannabinoid CB1 receptor that THC uses. Trial authors link its calming effect to these pathways and note a bell-shaped dose response seen in animals and in one human trial; the mechanism in people is not established.[5][4]
What the human research says
Human trials only, one row per outcome. The form column shows what each trial actually tested. Grades follow our evidence-grading policy.
At a glance: strongest evidence per outcome
| Outcome | Grade | Effect | Strongest evidence | Rows |
|---|---|---|---|---|
| Anxiety | Grade C | Improved | Meta-analysis · n = 316 · 2024 | 5 rows |
| Sleep | Grade D | Mixed | Pilot RCT · n = 30 · 2024 | 1 row |
Every graded row
| Outcome | Effect | Grade | Best evidence | PMID |
|---|---|---|---|---|
| Anxiety symptoms across anxiety disorders | Improved | Grade C | Meta-analysis · n = 316 · 2024 People with generalized anxiety, social anxiety or post-traumatic stress disorder (8 studies) Form: Oral CBD · Dose: Varied across studies · Duration: Varied across studies Pooled analysis found a large reduction in anxiety with CBD (Hedges g -0.92, 95% CI -1.80 to -0.04); the authors urge caution because the samples were small. | 38924898 (opens PubMed)Source [8] |
| Anxiety and sleep scores in a psychiatric clinic | Improved | Grade D | Observational · n = 72 · 2019 Adult psychiatric outpatients with anxiety or poor sleep (US) Form: CBD capsules added to usual care · Dose: Nearly all patients 25 mg/day (range 25–175 mg/day) · Duration: Monthly follow-up Anxiety scores fell in the first month in 79% of patients; sleep scores improved in 67% but fluctuated over time. | 30624194 (opens PubMed)Source [14] |
| Anxiety during a simulated public speaking test at different doses | Mixed | Grade D | RCT · n = 57 · 2019 Healthy men Form: Oral CBD · Dose: 150, 300 or 600 mg once · Duration: Single dose Only 300 mg reduced anxiety during the speech versus placebo; 150 mg and 600 mg did not differ from placebo. | 30328956 (opens PubMed)Source [5] |
| Anxiety before a cancer scan | Mixed | Grade D | RCT · n = 50 · 2024 Women with advanced breast cancer and clinical anxiety (US) Form: Purified prescription CBD oral solution · Dose: 400 mg once · Duration: Single dose The primary outcome (change in fear score) did not differ from placebo; fear scores 2–4 hours after the dose were lower with CBD. No grade 3 or 4 side effects. | 39680411 (opens PubMed)Source [6] |
| Insomnia severity and sleep measures | Mixed | Grade D | Pilot RCT · n = 30 · 2024 Adults with primary insomnia (Australia) Form: CBD oil under the tongue · Dose: 150 mg nightly, 60 minutes before bed · Duration: 2 weeks Insomnia severity, time to fall asleep and time awake did not differ from placebo; well-being scores and actigraph sleep efficiency at week 2 were better with CBD. | 38174873 (opens PubMed)Source [7] |
| Anxiety during a simulated public speaking test in social anxiety disorder | Improved | Grade D | RCT · n = 24 · 2011 Never-treated adults with generalized social anxiety disorder (Brazil) Form: Oral CBD · Dose: 600 mg once, 90 minutes before the test · Duration: Single dose CBD reduced anxiety, cognitive impairment and discomfort during the speech compared with placebo. | 21307846 (opens PubMed)Source [4] |
Forms
| Form | Elemental % | Absorption (human data) | GI tolerance | Studied for |
|---|---|---|---|---|
| Purified CBD oral solution (Epidiolex, prescription) | Varies[3][6] | A high-fat meal raised peak levels about 5-fold and total exposure about 4-fold compared with fasting.[3] | Common side effects in epilepsy trials: sleepiness, decreased appetite, diarrhea, liver enzyme rises, fatigue, rash and sleep problems.[3] | Seizures in Lennox-Gastaut syndrome, Dravet syndrome and tuberous sclerosis complex; used in the scan-anxiety trial |
| Over-the-counter CBD oils, capsules and gummies | Varies[2][7][10] | No absorption data for these products in our sources.[2] | NCCIH notes these products may contain more or less CBD than the label states and may contain THC or other contaminants.[2] | Insomnia pilot (150 mg oil) and exercise safety trial (60 mg oil or solubilisate) |
Studied doses
Doses studied in human trials ranged from 60 mg/day for a week in an exercise safety trial to 1,500 mg/day for about 3.5 weeks in a liver safety study. Anxiety trials used single doses of 150–600 mg, and the insomnia pilot used 150 mg nightly for 2 weeks.[10][9][4][5][7]
Interactions
Published interactions only: each row cites a source. No row means none was found in our sources, not that a combination is safe.
Medicines
- TacrolimusPK studyCBD titrated to 5 mg/kg twice daily raised tacrolimus peak levels 4.2-fold and total exposure 3.1-fold in healthy adults; the authors advise dose reduction and frequent level checks.[11]
- Caffeine and other CYP1A2 substratesPK studySteady-state CBD (750 mg twice daily) raised caffeine exposure 95% and doubled its half-life; the label advises considering lower doses of CYP1A2 substrates.[12][3]
- Valproate and clobazamDrug labelRaise the risk of liver enzyme elevations with prescription CBD; ALT above 3 times normal occurred in 30% of epilepsy patients on both drugs versus 3% on neither.[3]
- Drugs metabolized by CYP2C8, CYP2C19, CYP2B6, UGT1A9 or P-gpDrug labelThe Epidiolex label advises considering dose changes for these drugs and a lower starting dose of oral everolimus.[3]
- Alcohol and other sedating drugsDrug labelMay add to sleepiness and sedation.[3]
Who should be careful
- Liver disease or drugs that stress the liver
- Prescription CBD can raise liver enzymes, especially with valproate, and 5 of 16 healthy adults on 1,500 mg/day met drug-induced liver injury criteria; liver tests are monitored with the prescription product.[3][9]
- People on prescription drugs with narrow dosing margins
- CBD raised tacrolimus exposure about 3-fold and changes levels of other drugs, so prescribers need to know about any CBD use.[11][3]
- Pregnancy and breastfeeding
- The Epidiolex label reports no adequate human data and developmental toxicity in pregnant animals.[3]
- Children
- Outside prescription Epidiolex, the FDA says no cannabis-derived product has been approved as safe and effective for children.[1]
- Drug-tested workers and athletes
- Over-the-counter CBD products may contain THC that is not on the label.[2]
- Driving
- The prescribing information warns of sleepiness and advises against driving until people know how CBD affects them.[3]
Talk to your pharmacist or clinician before starting a supplement, especially if you take prescription medicines.
Safety first
CBD has real side effects and drug interactions, and products sold as supplements are outside the FDA framework.
- Liver enzymes
- In a phase I study, 7 of 16 healthy adults on 1,500 mg/day had ALT above normal, and all elevations began within 2–4 weeks; 6 were withdrawn from the study.[9]
- Lower doses
- A 7-day crossover trial of 60 mg/day in 17 trained athletes found no rise in liver enzymes compared with placebo.[10]
- Side effects
- In epilepsy trials, CBD raised the risk of serious adverse events (RR 2.67) and of stopping because of side effects (RR 3.95) versus control.[13]
- Product quality
- NCCIH notes that over-the-counter products may be mislabeled or contaminated with THC.[2]
How to read the label
- Label dose vs clinical doseAnxiety trials used single doses of 300–600 mg and the insomnia pilot 150 mg; prescription dosing is 5–25 mg/kg/day (about 350–1,750 mg for a 70 kg adult).[5][4][7][3]
- Label accuracyNCCIH warns that over-the-counter CBD products may contain more or less CBD than stated, and may contain THC.[2]
- Legal status of the productThe FDA says CBD cannot be sold legally as a dietary supplement or added to food in interstate commerce; Epidiolex is the only FDA-approved CBD drug.[1]
Cost per studied dose, without the sales pitch
Compare any product you already have by the amount it delivers, not by capsule count or front-of-pack weight.
Cost per studied dose
Enter the price, servings and the amount per serving from your label.
Frequently asked questions
Does CBD help with anxiety?
Small trials found single 300–600 mg doses lowered anxiety during a public speaking test, and a pooled analysis of 8 studies found a benefit with a wide margin of uncertainty. A 400 mg dose did not meet its main goal in a cancer-scan anxiety trial.[4][5][8][6]
Does CBD help you sleep?
In a 2-week pilot in 30 adults with insomnia, 150 mg nightly did not change insomnia severity or time to fall asleep, though sleep efficiency on wrist monitors was better at week 2.[7]
Can CBD harm the liver?
It can. At 1,500 mg/day, 5 of 16 healthy adults had liver enzyme rises meeting drug-injury criteria, and the Epidiolex label calls for liver monitoring. A week of 60 mg/day did not raise liver enzymes in athletes.[9][3][10]
Does CBD interact with medications?
Yes. It raised tacrolimus levels about 3-fold and doubled the half-life of caffeine, and the Epidiolex label lists interactions with valproate, clobazam and several drug-metabolizing enzymes.[11][12][3]
Is CBD legal as a supplement?
The FDA says CBD products cannot be sold legally as dietary supplements or added to food sold across state lines. State rules vary.[1][2]
Will CBD make me fail a drug test?
It can if the product contains THC; NCCIH warns that over-the-counter products may contain THC not listed on the label.[2]
Is CBD safe during pregnancy?
There are no adequate human data, and animal studies of CBD showed developmental toxicity, according to the Epidiolex label.[3]
Sources
- [1]FDA Regulation of Cannabis and Cannabis-Derived Products, Including Cannabidiol (CBD): Questions and Answers (content current as of July 16, 2024). U.S. Food and Drug Administration, 2024. fact-sheet
- [2]Cannabis (Marijuana) and Cannabinoids: What You Need To Know (last updated November 2019). National Center for Complementary and Integrative Health (NCCIH), 2019. fact-sheet
- [3]EPIDIOLEX (cannabidiol) oral solution: prescribing information (revised July 2025). Jazz Pharmaceuticals via DailyMed (U.S. National Library of Medicine), 2025. label
- [4]Cannabidiol reduces the anxiety induced by simulated public speaking in treatment-naïve social phobia patients.. Neuropsychopharmacology, 2011. RCT
- [5]Cannabidiol presents an inverted U-shaped dose-response curve in a simulated public speaking test.. Braz J Psychiatry, 2019. RCT
- [6]Cannabidiol for Scan-Related Anxiety in Women With Advanced Breast Cancer: A Randomized Clinical Trial.. JAMA Netw Open, 2024. RCT
- [7]Cannabidiol for moderate-severe insomnia: a randomized controlled pilot trial of 150 mg of nightly dosing.. J Clin Sleep Med, 2024. pilot-RCT
- [8]Therapeutic potential of cannabidiol (CBD) in anxiety disorders: A systematic review and meta-analysis.. Psychiatry Res, 2024. meta-analysis
- [9]Cannabidiol and Abnormal Liver Chemistries in Healthy Adults: Results of a Phase I Clinical Trial.. Clin Pharmacol Ther, 2021. open-label
- [10]Short-term repeated oral intake of low dose cannabidiol: effects on liver enzyme activity and creatinine concentration during intense exercise.. Arch Toxicol, 2025. crossover-RCT
- [11]A Phase I Trial of the Pharmacokinetic Interaction Between Cannabidiol and Tacrolimus.. Clin Pharmacol Ther, 2025. PK-study
- [12]A Phase 1 Open-Label, Fixed-Sequence Pharmacokinetic Drug Interaction Trial to Investigate the Effect of Cannabidiol on the CYP1A2 Probe Caffeine in Healthy Subjects.. Clin Pharmacol Drug Dev, 2021. PK-study
- [13]Adverse Events of Cannabidiol Use in Patients With Epilepsy: A Systematic Review and Meta-analysis.. JAMA Netw Open, 2023. meta-analysis
- [14]Cannabidiol in Anxiety and Sleep: A Large Case Series.. Perm J, 2019. observational
What changed
- First draft. Every PMID checked against NCBI E-utilities esummary (title and year taken from the record) and abstracts read for the cited rows.
- Adversarial review pass: every evidence row re-read against its PubMed abstract. Official sources opened before citing: FDA cannabis and CBD questions and answers (content current July 16, 2024; dietary supplement exclusion, children), NCCIH cannabis and cannabinoids fact sheet (November 2019; label accuracy, THC contamination, side effects) and the Epidiolex prescribing information on DailyMed (revised July 2025; liver enzymes, interactions, food effect, pregnancy). Industry ties: GW Research consultants and Greenwich Biosciences employees authored the 1,500 mg/day liver study; GW Research sponsored the caffeine study; Cannvalate funded the insomnia pilot; STI-Pharm supplied CBD and authors hold a fluorinated-CBD patent in the dose-response trial (coi.py / PubMed statements). Kept out: 2025 JAMA Internal Medicine liver-enzyme RCT (40622698, has an erratum) and the 2018 phase I safety trial (30374683, has an erratum). Retraction, expression-of-concern, erratum and withdrawn checks on all 11 PMIDs: none flagged. Safety-first page with no product links. Page set indexable. Independent browser review (4 parallel reviewers, every evidence row checked against its abstract and official sources on the Vercel preview) found: no factual errors; source dates confirmed.