CBD for Anxiety: Independent Evidence on Sleep, Liver Risk & Law

Table of contents

• Evidence summary
• What CBD is, and its legal status
• How it works
• What works and what does not
• Benefits by claim
• Risks and all side effects
• All interactions
• Who should avoid CBD
• Doses studied and product quality
• Follow the money: who makes it and who funded the evidence
• Related research
• Frequently asked questions
• Sources and funding notes

Evidence summary

Independent evidence and credibility scorecard

Independence tiers: 1 = regulator or government body; 2 = academic or Cochrane team with no declared commercial funding for the work; 3 = independent authors analysing mostly manufacturer-run trials, or minor declared industry ties; 4 = manufacturer-funded or manufacturer-authored. Credibility runs from A (highest) to D.

Note on clustering: most positive anxiety experiments come from one university in Brazil. The safety and regulatory evidence comes mainly from the US, UK and EU. Independent trials from the Netherlands and the UK were negative.

What CBD is, and its legal status

Cannabidiol is one of more than 80 cannabinoids in the cannabis plant. Unlike THC, it does not cause a "high". In a human abuse-potential study reported on the label, single doses of 750 to 4,500 mg in recreational drug users gave "drug liking" scores within the placebo range, and Epidiolex is not a controlled substance in the US (Epidiolex label).

Two very different products share one name

• Prescription CBD (Epidiolex, sold as Epidyolex in Europe): a purified oral solution of 100 mg/mL, approved for seizures in Lennox-Gastaut syndrome, Dravet syndrome and tuberous sclerosis complex from age 1. Doses run from 5 to 25 mg per kg of body weight a day, with liver tests before and during treatment (Epidiolex label). For a 70 kg adult, 10 mg/kg is 700 mg a day.
• Consumer CBD: oils, drops, capsules, gummies, drinks, vapes and creams made from hemp. In the US, the FDA says no CBD product other than Epidiolex is approved and it has sent warning letters to firms making treatment claims (FDA Q&A; warning-letter list). Contents vary widely (see product quality).

Legal status by country

How it works

No one has shown in humans how CBD might reduce anxiety. The main ideas are:

• Serotonin 5-HT1A receptors (laboratory only): in cultured cells, CBD acted as a "modest affinity" activator of the human 5-HT1A receptor, the same target as the anxiety drug buspirone (Russo 2005).
• The body's own cannabinoid, anandamide (some human data): CBD slows the breakdown of anandamide. In a German trial in schizophrenia, CBD raised blood anandamide levels (Leweke 2012). The Dutch anxiety trial was built on this idea and found no clinical benefit (Kwee 2022).
• Brain blood flow (one tiny human study): in 10 people with social anxiety, 400 mg CBD lowered subjective anxiety and changed blood flow in limbic brain areas (Crippa 2011). A UK brain-scan study of 600 mg in 24 healthy people found no effect on emotional processing or induced anxiety (Bloomfield 2022).
• An unusual dose curve: in Brazilian experiments, only a middle dose (300 mg) worked, and higher doses did not (Zuardi 2017; Linares 2019). A Dutch systematic review of 87 human and animal studies found CBD had no effect on anxiety in 70.3% of observations and no clear link between blood levels and effect (Kwee 2022 review; no conflicts declared).

What works and what does not

Benefits by claim

Does CBD work for anxiety?

The honest answer is that it has not been properly tested where it matters. The independent Australian review by Black and colleagues searched every cannabinoid study up to April 2018. Of the anxiety trials it included, most tested THC-containing drugs in people with chronic pain or multiple sclerosis. Few trials tested CBD alone. The authors concluded that the evidence for cannabinoids in anxiety disorders is weak and insufficient to guide use (Black 2019). In the small number of studies of people with an anxiety disorder, they saw no significant benefit from CBD (full text).

A 2025 meta-analysis from the University of São Paulo group reviewed cannabinoid formulations for anxiety-related disorders and Tourette syndrome. The authors rated the evidence on pharmaceutical-grade CBD as low quality and said CBD "may have limited efficacy" for social anxiety and GAD (Raminelli 2025). We do not quote its pooled CBD figures because we could not re-check them. The studies are small, several come from the same team, and we could not read the paper's funding statement. Members of this group hold a CBD-compound patent (Linares 2019 disclosure).

Social anxiety and public speaking

This is where CBD's reputation comes from. The research model asks people to prepare and give a short speech while their anxiety is rated.

• In 1993, 300 mg CBD reduced anxiety after a simulated speech in 40 healthy volunteers (10 per group); diazepam did too (Zuardi 1993).
• In 2011, 24 untreated people with social anxiety disorder took 600 mg CBD or placebo 90 minutes before speaking. The CBD group had less anxiety, less discomfort and fewer negative thoughts about themselves. Their scores came close to those of 12 healthy people who took nothing (Bergamaschi 2011).
• In 2017, 60 healthy volunteers spoke in front of each other. Anxiety after the speech was lower with 300 mg CBD and with clonazepam, but not with 100 mg or 900 mg CBD (Zuardi 2017).
• In 2019, 57 healthy men took 150, 300 or 600 mg. Only 300 mg reduced anxiety during the speech (Linares 2019).
• A UK team gave 600 mg to 24 healthy people and found no effect on experimentally induced anxiety (Bloomfield 2022). Its funding statement could not be re-checked.

For daily use, a Japanese pilot gave 300 mg a day for 4 weeks to 18- and 19-year-olds with social anxiety and avoidant personality disorder. Scores on the Fear of Negative Evaluation scale fell on CBD and barely moved on placebo (Masataka 2019). It is encouraging but small: 37 people were analysed (17 on CBD, 20 on placebo), and the result comes from a single author.

The strongest test so far was negative. In the Netherlands, 80 people with panic disorder or social anxiety who had not responded to earlier treatment took 300 mg CBD or placebo before each of 8 exposure-therapy sessions. CBD did not improve outcomes at the end of treatment or at 6-month follow-up (Kwee 2022).

Everyday stress and burnout

During the COVID-19 pandemic, the São Paulo group randomised 120 hospital staff to 300 mg CBD a day plus usual care, or usual care alone, for 28 days. Emotional exhaustion scores were lower on CBD than on usual care alone at day 28, though the size of the difference could not be re-checked. There was no placebo, so staff knew what they were taking, and 5 people on CBD had serious adverse events, all of which resolved after stopping treatment (Crippa 2021). Funding and author disclosures for this trial were not re-checked.

A US psychiatric clinic reviewed records for 72 adults given CBD alongside usual care. Anxiety scores fell in the first month in 57 (79.2%) (Shannon 2019). With no comparison group, this cannot separate CBD from other treatment, time or expectation.

Does CBD help you sleep?

The evidence is thin. The US National Center for Complementary and Integrative Health notes some studies report better sleep with cannabinoids but says it is unclear whether the products caused it (NCCIH).

• In 30 Australian adults with insomnia, 150 mg CBD nightly for 2 weeks did not change insomnia severity, time to fall asleep, sleep efficiency by diary, or night waking. CBD users reported better well-being, and wrist-sensor sleep efficiency was better at week 2 (Narayan 2024). Daytime thinking was unaffected (Narayan 2025). Both papers were funded by, or used product supplied by, CBD companies.
• In 27 healthy volunteers, 300 mg CBD before bed had no effect on laboratory sleep recordings (Linares 2018).
• Somnolence affected 32% of epilepsy patients on prescription CBD against 11% on placebo, yet insomnia and poor sleep were also listed as common side effects (Epidiolex label).

Risks and all side effects

In epilepsy patients, the risk depends heavily on other medicines. The chart below uses the same threshold (ALT more than 3 times normal) across studies.

What regulators say about long-term risk

The FDA says it has not found adequate evidence on how much CBD can be taken, and for how long, before it causes harm (FDA 2023). The UK FSA says the more CBD people take over their lifetime, the more likely long-term effects such as liver damage or thyroid problems become (FSA). The FDA trial above found no change in thyroid hormones or testosterone over 4 weeks (Florian 2025).

Male fertility (animal data)

In laboratory animals, CBD caused smaller testes, impaired sperm development and lower testosterone, including in male offspring of treated mothers. The FDA says it is not yet clear what this means for people (FDA consumer update).

Other side effects

• Common with prescription CBD (10% or more, and more than placebo): sleepiness; reduced appetite; diarrhoea; raised liver enzymes; tiredness and weakness; rash; insomnia and poor sleep; infections (Epidiolex label).
• Sedation: 46% of patients also taking clobazam were sleepy or sedated, against 16% not taking it. The label advises against driving until people know how it affects them (Epidiolex label).
• Mood: the Epidiolex label carries a warning to watch for suicidal thoughts and behaviour (Epidiolex label).
• Allergy: hypersensitivity-type reactions were reported in the high-dose Watkins 2021 study (Watkins 2021).
• Lower doses: dry mouth was more common on 150 mg than placebo in the insomnia trial (Narayan 2025).
• Unexpected THC: mislabelled products can contain THC, which can cause intoxication or a positive drug test (Bonn-Miller 2017).

All interactions

Most of these interactions were measured at prescription doses. Whether lower consumer doses cause the same effects has not been well studied, which is why the UK FSA advises anyone taking medication to speak to a health professional first (FSA).

Who should avoid CBD

• Pregnancy, trying to conceive, and breastfeeding: the FSA says CBD is not recommended in these groups as a precaution unless under medical direction (FSA). In animals, CBD caused developmental harm, and there are no data on CBD in human breast milk (Epidiolex label).
• Children and teenagers under 18: not recommended unless under medical direction (FSA). The FSA also lists children as a vulnerable group.
• Anyone taking prescription medicines: especially anti-seizure drugs, blood thinners, sedatives and drugs listed above (FSA; Epidiolex label).
• Liver disease: moderate or severe liver impairment requires lower prescription doses (Epidiolex label).
• Drivers and machine operators: until they know whether CBD makes them drowsy (Epidiolex label).
• People subject to drug testing: some products contain undeclared THC (Spindle 2022).
• People with a diagnosed anxiety disorder: CBD has no proven role in treatment; effective options are covered in our related guides below.

Doses studied and product quality

These are doses used in research or stated by regulators, not recommendations.

This creates a gap that sellers rarely mention. The doses that showed any effect on anxiety (300–600 mg) are 30 to 60 times the UK advisory limit. At the FSA's example strength, 300 mg would be well over 100 drops of 5% oil a day. At those doses, liver-enzyme rises become a measurable risk (Florian 2025). Low-dose shop products have not been shown to reduce anxiety in any trial we found.

Product quality problems

• Online extracts (US, 2017): of 84 products from 31 companies, 31% were within 10% of the labelled CBD, 43% contained more and 26% contained less. THC was found in 18 (21%). Vape liquids were worst, with 88% mislabelled (Bonn-Miller 2017).
• Topicals (US, 2022): of 89 products stating a CBD amount, 24% were accurate. THC was found in 37 of 105 products (35%), including 4 labelled "THC free". 28% made therapeutic claims (Spindle 2022).
• FDA sampling (2020): of 147 products tested, 49% contained THC and only 45% of those declaring a CBD amount were within 20% of it, as summarised by a law firm from the FDA's report to Congress (Covington 2020).
• The NIH notes that over-the-counter CBD may contain more or less CBD than stated (NCCIH).

Follow the money: who makes it and who funded the evidence

Who makes and sells it

• Jazz Pharmaceuticals / GW Pharmaceuticals: GW developed Epidiolex; its research arm, GW Research Ltd, is based in Cambridge, UK (Watkins 2021). Jazz completed its purchase of GW on 5 May 2021 for about $7.2 billion ($6.7 billion net of GW's cash), paying $220 per GW ADS, each representing 12 ordinary shares (Jazz SEC filing; completion notice). Epidiolex/Epidyolex net sales rose 9% to $1.1 billion in 2025 (Jazz results, February 2026). Prescription CBD is a large business on its own, and it is approved only for epilepsy.
• Consumer CBD brands: thousands of mostly small firms sell CBD oils and gummies. We found no market-size figure from an independent or government source, so we have not quoted one.
• CBD suppliers to research: STI-Pharm (UK) and BSPG-Pharm (UK) supplied free CBD to the São Paulo trials (Linares 2019; Zuardi 2017). The supplier for the UK brain-scan study could not be re-checked (Bloomfield 2022). Brains Bioceuticals (UK) supplied the insomnia trial product (Narayan 2024).

Who funded the trials

• The Brazilian public-speaking trials: funded by Brazil's national research council (CNPq) and São Paulo's research foundation. The disclosures in later papers state that several senior authors co-invented fluorinated CBD compounds, that the University of São Paulo licensed this patent to Phytecs, and that the university has an agreement with Prati-Donaduzzi to develop a CBD medicine for conditions including anxiety disorders (Linares 2019). The 2011 social-anxiety study was funded by FAPESP, and its CBD was supplied free by STI-Pharm and THC-Pharm; it did not mention the patent, which was filed later (Bergamaschi 2011 full text).
• The negative Dutch trial: ZonMw (the Dutch health research council) and the Dutch Brain Foundation; sponsor Utrecht University (Kwee 2022).
• The Japanese trial: Japan Society for the Promotion of Science grant; no commercial ties declared (Masataka 2019).
• The insomnia trial: Cannvalate, an Australian medical-cannabis company, and the Barbara Dicker Brain Sciences Foundation; product from Brains Bioceuticals. The authors state the sponsors provided funding only (Narayan 2024). The companion cognition paper acknowledges funding from Brains Bioceuticals (Narayan 2025).
• The liver studies: one was run and paid for by the FDA (Florian 2025). The other was sponsored by GW Research, (Watkins 2021).
• The epilepsy trials: funded by GW Pharmaceuticals (Devinsky 2017).
• The main independent review: Australian government bodies and the US NIH (Black 2019). NCCIH, which publishes the NIH fact sheet, is funded by US Congress (appropriations).

Documented integrity events

• FTC action: on 17 December 2020, the US Federal Trade Commission acted against six sellers for unsupported claims that their products could treat serious conditions including cancer, heart disease, high blood pressure and Alzheimer's disease (FTC).
• FDA warning letters: the FDA publishes a running list of warning letters to firms selling CBD with unapproved drug claims, including during the COVID-19 pandemic (FDA list).
• UK retailer withdrawal: after the FSA's October 2023 advice, Holland & Barrett removed 31 CBD products from sale (NutraIngredients 2023).
• Mislabelling: repeated independent tests show inaccurate CBD content and undeclared THC (see product quality).

Related research

For wider, evidence-graded context, see these Pure City Research guides:

• Generalised anxiety disorder: treatments, supplement evidence and safety
• Social anxiety disorder: assessment, treatment evidence and safety
• Supplements for stress, anxiety and depression: the evidence
• Sleep supplements: the evidence
• Insomnia
• CBT-I: cognitive behavioural therapy for insomnia
• Drug and supplement induced liver injury
• Anticoagulants and bleeding safety
• Oral lavender (Silexan)
• L-theanine
• Propranolol for anxiety
• The COA transparency audit: supplement lab results

Frequently asked questions

Sources and funding notes

What we could not verify: the funding statement and pooled CBD figures of the 2025 Raminelli meta-analysis (the journal page was paywalled or blocked, so we quote only its stated conclusions); the CBD supplier in the Japanese trial; the funding of the 2022 cocaine-use-disorder trial; the funding statements of the Masataka 2019, Bloomfield 2022, Crippa 2021 and Kwee 2022 trials (the Kwee protocol page was blocked, so the ZonMw and Dutch Brain Foundation attribution is not confirmed); the exact effect sizes for Masataka 2019, Crippa 2021 and Kwee 2022; the arm sizes of the FDA liver trial (summaries give 141 or 151 CBD participants, so the 5.6% figure is quoted as published in summaries); the Mongeau-Pérusse, Shannon 2019 and Kwee 2022 review figures; the Black 2019 anxiety subgroup numbers, its funding and search dates; the Narayan 2025 funding note; the Bergamaschi 2011, Linares 2019 and Bloomfield 2022 supplier and patent details, which come from the papers' disclosures and were not re-checked; the Epidiolex clobazam sedation percentages; the FSA statement on liver and thyroid risk; and the EFSA 2022 count of 19 paused applications. The NIH LiverTox page on cannabidiol and the Australian TGA's page on over-the-counter low-dose CBD were blocked by bot checks, so we did not use them. We could not open the FDA's 2020 sampling report directly and relied on a law-firm summary. We found no independent figure for the size of the consumer CBD market. We could not confirm how the November 2025 US hemp law, now delayed in part to 11 December 2026, will apply to full-spectrum CBD products, and we could not open primary government sources for reports of a 2026 Medicare pilot covering CBD or of FDA enforcement policy linked to it, so we have not described them. We found no large, long, placebo-controlled trial of CBD alone in generalised anxiety disorder.

1. FDA statement: existing frameworks not appropriate for CBD (26 January 2023) — US regulator.
2. FDA consumer update on cannabis and CBD products — US regulator.
3. FDA regulation of cannabis-derived products: Q&A — US regulator.
4. FDA warning letters and test results for CBD-related products — US regulator.
5. Florian J et al. Cannabidiol and liver enzyme level elevations in healthy adults. JAMA Intern Med 2025 (full text) — randomised trial; FDA-funded, lead author at FDA.
6. Epidiolex (cannabidiol) prescribing information, DailyMed — FDA-approved label; maker Jazz Pharmaceuticals.
7. Food Standards Agency: Cannabidiol (CBD) consumer advice — UK regulator.
8. EFSA: Cannabidiol novel food evaluations on hold pending new data (June 2022) — EU food-safety agency.
9. NCCIH: Cannabis (marijuana) and cannabinoids — US government fact sheet; funded by Congress (appropriations history).
10. Congressional Research Service IN12620: change to federal definition of hemp (December 2025) — US Congress research service.
11. Congressional Research Service IF13136: changes to the statutory definition of hemp (updated 28 September 2026) — US Congress research service; reports the delay to 11 December 2026.
12. Executive Order 14370: Increasing medical marijuana and cannabidiol research (18 December 2025) — US presidency.
13. EFSA: provisional safe level for cannabidiol as a novel food (9 February 2026) — EU food-safety agency.
14. FTC: crackdown on deceptively marketed CBD products (17 December 2020) — US regulator.
15. Black N et al. Cannabinoids for the treatment of mental disorders: systematic review and meta-analysis. Lancet Psychiatry 2019 (full text) — funded by the Australian TGA, Department of Health, NHMRC and US NIH; some authors held untied industry grants.
16. Raminelli AO et al. Cannabinoid formulations for anxiety-related disorders and Tourette syndrome: meta-analysis. Cannabis Cannabinoid Res 2025 — University of São Paulo group; funding statement not accessed.
17. Kwee CM et al. CBD in clinical and preclinical anxiety research: concentration-effect review. J Psychopharmacol 2022 — Dutch academic; no conflicts declared.
18. Bergamaschi MM et al. CBD reduces anxiety induced by simulated public speaking in social phobia. Neuropsychopharmacology 2011 (full text) — FAPESP grant; CBD supplied free by STI-Pharm and THC-Pharm.
19. Zuardi AW et al. Effects of ipsapirone and cannabidiol on human experimental anxiety. J Psychopharmacol 1993 — Brazilian academic; funding not stated in record.
20. Zuardi AW et al. Inverted U-shaped dose-response of CBD during public speaking. Front Pharmacol 2017 — CNPq grant; CBD supplied free by BSPG-Pharm; authors' patent licensed to Phytecs.
21. Linares IM et al. CBD inverted U-shaped dose-response in a simulated public speaking test. Braz J Psychiatry 2019 — CBD supplied free by STI-Pharm; authors' patent licensed to Phytecs; university agreement with Prati-Donaduzzi.
22. Crippa JA et al. Neural basis of anxiolytic effects of CBD in social anxiety disorder. J Psychopharmacol 2011 — Brazilian academic; funding not accessed.
23. Masataka N. Anxiolytic effects of repeated CBD treatment in teenagers with social anxiety disorders. Front Psychol 2019 — Japanese government grant (JSPS); no conflicts declared.
24. Kwee CM et al. CBD enhancement of exposure therapy in treatment-refractory social anxiety and panic disorder: RCT. Eur Neuropsychopharmacol 2022 (full text) — ZonMw and Dutch Brain Foundation; no conflicts.
25. Bloomfield MAP et al. Acute effects of CBD on emotional processing and anxiety. Psychopharmacology 2022 — funding and disclosures not re-checked.
26. Mongeau-Pérusse V et al. CBD effect on anxiety in cocaine use disorder: exploratory RCT results. J Addict Med 2022 — Canadian academic; no conflicts declared; funding not checked.
27. Crippa JAS et al. CBD for burnout in frontline health care workers: randomised clinical trial. JAMA Netw Open 2021 — funding and disclosures not re-checked.
28. Shannon S et al. Cannabidiol in anxiety and sleep: a large case series. Perm J 2019 — clinic chart review; no product funding declared.
29. Narayan AJ et al. CBD for moderate-severe insomnia: randomised pilot of 150 mg. J Clin Sleep Med 2024 — Cannvalate and Barbara Dicker Brain Sciences Foundation; product from Brains Bioceuticals.
30. Narayan AJ et al. Nightly 150 mg CBD and daytime cognition in insomnia. Psychopharmacology 2025 — funding and product from Brains Bioceuticals.
31. Linares IMP et al. No acute effects of CBD on the sleep-wake cycle of healthy subjects. Front Pharmacol 2018 — Brazilian academic.
32. Watkins PB et al. CBD and abnormal liver chemistries in healthy adults: phase I trial. Clin Pharmacol Ther 2021 — reported as sponsored by GW Research; author affiliations not re-checked.
33. Grayson L et al. An interaction between warfarin and cannabidiol: case report. Epilepsy Behav Case Rep 2018 — single-patient case report; author affiliations not re-checked.
34. Devinsky O et al. Trial of cannabidiol for drug-resistant seizures in Dravet syndrome. N Engl J Med 2017 — funded by GW Pharmaceuticals.
35. Russo EB et al. Agonistic properties of cannabidiol at 5-HT1a receptors. Neurochem Res 2005 — cell study; funding not checked.
36. Leweke FM et al. CBD enhances anandamide signalling and alleviates psychotic symptoms. Transl Psychiatry 2012 — German academic trial; funding not checked.
37. Bonn-Miller MO et al. Labeling accuracy of cannabidiol extracts sold online. JAMA 2017 (funding note) — Institute for Research on Cannabinoids; authors' fees from cannabis companies declared.
38. Spindle TR et al. Cannabinoid content and label accuracy of hemp-derived topical products. JAMA Netw Open 2022 (full text) — US SAMHSA; authors' consulting ties declared.
39. Covington & Burling: FDA submits report on CBD marketplace surveillance to Congress (July 2020) — law-firm summary of an FDA report.
40. Jazz Pharmaceuticals second-quarter 2021 results (SEC filing); GW acquisition completion notice (May 2021) — company filings.
41. Jazz Pharmaceuticals full-year 2025 results (24 February 2026) — company press release.
42. NutraIngredients: Holland & Barrett pulls 31 CBD products (October 2023) — trade press.

Educational research reviewed 6 October 2026. This article is not medical advice; talk to a qualified clinician or pharmacist before starting, stopping or combining treatments.