Does CBD work? What the clinical studies actually say in 2026

"Does it work?" is the question most frequently asked about CBD. The honest answer is neither yes nor no, it is a map: there are validated clinical indications, avenues that have been studied without being conclusive, and wellbeing uses that rest on individual testimony without scientific validation.

The only use of CBD validated by the European (EMA) and French agencies (ANSM) is Epidyolex (cannabidiol as an oral solution), a hospital medicine available on prescription. It is authorised as an adjunctive treatment for certain drug-resistant epilepsies: Lennox-Gastaut syndrome, Dravet syndrome, and tuberous sclerosis complex.

A French multicentre study published in 2025 in Epilepsia assessed the real-world effectiveness and the safety at 18 months of CBD in drug-resistant epileptic patients. The results confirm a reduction in seizures in a share of the patients, with an acceptable tolerance profile but one marked by adverse effects (drowsiness, raised transaminases, reduced appetite).

This is currently the only setting in which the effectiveness of CBD is documented at a level of clinical evidence acceptable to regulators. And it is not a wellbeing use: it is a medicine, prescribed in a hospital setting, at very high therapeutic doses (up to 20 mg/kg/day).

Anxiety

Several studies have explored the anxiolytic effect of CBD at high doses:

• Bergamaschi et al. 2011: randomised trial in 24 patients with social phobia, a single 600 mg dose of CBD before a public-speaking test, reduced anxiety compared with placebo. Small sample size.
• Crippa et al. 2011: brain imaging study, acute effect on the areas involved in anxiety.
• Hundal et al. 2018: no significant effect found at 600 mg in 32 paranoid patients.

The trials are few in number, based on small samples, at doses far above wellbeing use (typically the 25 mg sold in shops), and the results are heterogeneous.

Sleep

• Shannon et al. 2019 (Permanente Journal): retrospective observational study in 72 patients, self-reported improvement in sleep. No randomised control group, probable expectancy bias.
• Later randomised trials: divergent results, often non-significant on objective criteria (polysomnography).

Chronic pain

• Several trials of Sativex (a medicine combining THC and CBD in equal parts) show modest effectiveness in multiple sclerosis spasticity.
• On CBD alone, the literature is sparse and inconclusive.

Chronic stress and general wellbeing

No quality clinical study validates a specific effect of CBD on everyday stress in healthy adults.

The EFSA halted its assessment of CBD as a novel food in 2022 because of toxicological gaps, and published an update in 2026. The recurring methodological criticisms:

1. Small samples: most trials include fewer than 100 participants.
2. No randomised control group in the observational studies.
3. Non-standardised doses: from 25 to 600 mg depending on the study.
4. Short duration: most trials run for a few weeks, not several months.
5. Conflicts of interest: a significant number of studies are funded by CBD manufacturers.
6. Heterogeneous products: isolate, full spectrum, broad spectrum — the differences matter.
7. Lack of replication: few studies are reproduced in independent double-blind conditions.

This is why no health claim is entered in the European register for cannabidiol. In France, no brand may therefore state that CBD helps you sleep, reduces anxiety or relieves pain.

• ANSM: no marketing authorisation for wellbeing CBD. Only Epidyolex is authorised in France.
• HAS: no recommendation on CBD in sleep disorders or anxiety. The HAS was referred the question of cannabis-based medicines in March 2025, with conclusions expected during 2026.
• ANSES: a cautious position, proposing a 1B reprotoxic classification at European level, and flagging poisonings caused by contaminated products.

A few useful reflexes if you come across a study:

• Sample size: fewer than 100 participants means a lead, not proof.
• Type of study: observational (testimony) < cohort < randomised double-blind placebo-controlled trial.
• Conflicts of interest: the "Funding" and "Conflicts of interest" sections at the end of the paper.
• Replication: have the results been reproduced by an independent team?
• Status: is the study American, French, international? Published in which journal (peer-reviewed or not)?
• Year: CBD science moves fast, a 2015 study may be obsolete.

Three points to know before any use:

1. On 9 February 2026 the EFSA published a provisionally safe daily dose of 0.0275 mg/kg/day (approximately 2 mg/day for 70 kg).
2. Since 15 May 2026, the DGAL has been withdrawing unauthorised ingestible CBD products (Novel Food).
3. Wellbeing CBD is not medical cannabis: the latter is prescribed in a hospital setting for five precise indications, and its general roll-out was notified to the European Commission in March 2025.

CBD is not a medicine. In case of persistent pain, insomnia or anxiety, consult a healthcare professional.

• Only Epidyolex (epilepsy) benefits from European clinical validation.
• The studies on anxiety, sleep and pain are preliminary and insufficient according to the EFSA.
• No health claim is entered in the European register for CBD.
• Trials often use doses far above wellbeing use.
• Reading a study: sample size, conflicts of interest, type (randomised vs observational).
• Distinguish wellbeing CBD (freely sold) from medical cannabis (hospital prescription).

• EFSA Journal 2026 — Update statement CBD novel food — primary
• ANSM — Topic file on cannabis for medical use — primary
• ANSM — Progress report, Feb. 2024 (final year of the trial programme) — primary