Cannabidiol (CBD).

Compound, plant and product are different levels

CBD is one compound found in cannabis. It is not the same thing as the whole plant, hemp or a whole-plant extract. Products may contain different amounts of CBD, THC and other cannabinoids, and may contain terpenes, residual solvents or pesticides. “Natural” describes origin; it does not establish quality control or a health benefit.

A purified medicine studied as a defined substance is different from a retail extract. For a separate question, see CBD and sleep.13

What the approved medicine tells us

Epidiolex, called Epidyolex in some countries, is a prescription oral solution containing purified CBD. Evidence for it does not automatically transfer to a shop-bought oil, gummy, vape liquid, cream or full-spectrum extract.

European product information describes liver-function monitoring and cautions about interactions with medicines including valproate and clobazam. Approval concerns a particular product and use. It does not establish that CBD treats pain, anxiety, cancer, insomnia or general wellbeing. An incomplete explanation of precisely how it works does not erase the medicine’s trial evidence, but it limits claims about why it should help another condition. Formulation, exposure and other medicines matter.

The cited US label covers seizures associated with the three specified conditions from age one. EU product information sets a different starting age of two and specifies combination treatment. These are decisions for this particular medicine, not evidence for every CBD product.516

A clinical trial: seizures in Dravet syndrome

In a 2017 double-blind trial funded by GW Pharmaceuticals, 120 children and young adults with drug-resistant Dravet syndrome were randomly assigned purified oral CBD or a placebo. Both groups continued their existing antiseizure medicines.

Over 14 weeks, the median number of convulsive seizures per month fell from 12.4 to 5.9 in the CBD group and from 14.9 to 14.1 in the placebo group. The adjusted difference between groups in percentage change was −22.8 percentage points, with a 95% confidence interval from −41.1 to −5.4. These group results do not predict an individual’s response.

Serious adverse events were reported in 10 of 61 people receiving CBD and three of 59 receiving placebo. Eight versus one withdrew because of adverse events. Drowsiness and abnormal liver-test results were more common with CBD. Other medicines matter: many participants with drowsiness also took clobazam. Caregivers reported seizures, and noticeable side effects may have made it easier to guess the treatment. Fourteen weeks of comparison do not settle long-term safety, effects in other epilepsies or claims about retail CBD.76

Does CBD help anxiety?

Some research is promising, but it does not establish that shop-bought CBD treats anxiety. In a 2024 trial in India, 178 adults with mild to moderate anxiety disorders were assigned a particular oral CBD formulation or placebo. Researchers reported greater improvement in an anxiety questionnaire with CBD over the study period. Several authors disclosed employment or leadership roles at pharmaceutical companies.10

A different small trial gave 50 women with advanced breast cancer and clinical anxiety a single dose of prescription CBD or placebo before a cancer scan. Its main comparison—the change in anxiety score before and after treatment—was not statistically significant. An additional comparison of scores after treatment favoured CBD and needs further study.11

These studies involved different people, products and questions. Neither proves another product will work or be safe for an individual. Anxiety has many possible causes and established care options; a health professional can discuss those and possible interactions.

Non-intoxicating is not the same as harmless

FDA identifies liver injury, drowsiness, diarrhoea, appetite changes and medicine interactions as real safety concerns. The clinical programme supporting the approved medicine required medical monitoring. FDA also highlights uncertainty about long-term use, pregnancy and breastfeeding, developing brains, animal findings concerning male reproduction, and contamination or inaccurate labelling in some unapproved products. The severity and frequency of a risk depend on product, exposure, population and follow-up.45

In the European product information, liver transaminases and bilirubin are checked at defined points and later as clinically indicated. This is a feature of supervised use of a defined medicine, not a general monitoring protocol for every product sold as CBD. Interactions can change exposure to CBD or to another medicine, so a label-level safety statement cannot be reduced to “CBD is non-intoxicating.”5

For pregnancy and breastfeeding, the US FDA strongly advises avoiding CBD products because important safety questions remain. If you are pregnant, planning pregnancy or breastfeeding and use prescribed cannabidiol for seizures, contact the prescribing specialist promptly for an individual plan; do not stop seizure treatment on your own.16

Can you drive after taking CBD?

CBD can cause sleepiness or changes in alertness, and some CBD products also contain THC, which can impair driving. In a 2020 randomised crossover study, 26 healthy adults completed supervised road-driving tests after vaporising different cannabis preparations or placebo. The CBD-dominant preparation did not produce a statistically significant difference from placebo on the study’s lane-weaving measure, while THC-dominant and combined THC/CBD preparations increased weaving in the earlier test. The small study could not exclude a meaningful effect of CBD. Its preparations, doses and timing cannot establish safety for other oils, gummies, medicines or people. CBD does not reliably cancel THC-related impairment.

Do not drive if drowsy or impaired; ask a clinician or pharmacist about the exact product and other medicines. Alcohol and sedating medicines can increase drowsiness. Driving laws and drug-testing rules vary by jurisdiction, and there is no universal waiting period established by this study. For more on THC and driving, see the THC profile.12134

What did a liver-safety study in healthy adults find?

In a 2025 randomized trial led by US FDA researchers, 201 healthy adults were assigned a defined oral CBD solution or placebo for 28 days; the CBD group received 250–550 mg per day, depending on body weight. Eight people receiving CBD had a liver enzyme level more than three times the upper limit of normal, compared with none receiving placebo; the FDA estimated the event proportion at 5.6% versus 0%.

These blood-test changes were not accompanied by liver-related symptoms in the study, and enzyme levels returned to normal after stopping CBD. An enzyme increase is a safety signal, not by itself proof of liver failure. The study used a specific oral preparation and exposure in healthy adults, not a survey of retail oils or every dose; it does not establish an individual’s risk or long-term outcome. A clinician can discuss a specific product, symptoms and other medicines.89

Why retail evidence is difficult to interpret

Unapproved products can vary between batches and may not contain the amount stated on the label. A study of purified CBD in a controlled formulation may not describe a commercial extract. How a product is taken affects exposure, and a stated amount does not tell you the blood concentration. Small, short or open-label studies can suggest useful questions while remaining unable to establish an effect or estimate rare harms.41

What a retail label cannot tell you

A bottle may state an amount per serving without showing whether that amount is consistent throughout the batch, how much is absorbed or whether other cannabinoids or contaminants are present. The FDA has reported products with inaccurate cannabinoid content and warned that treatment claims can delay appropriate diagnosis or care.

This does not prove every unapproved product is ineffective. It does explain why evidence for a named prescription medicine cannot be assumed to apply to an unreviewed product.41

Can a CBD product contain THC?

Yes. THC is a different cannabinoid that can impair alertness, and its presence depends on the product. The surveys below cannot establish what is in a particular bottle or in today’s entire market. Product, batch, route and laboratory methods matter.

In a 2022 US laboratory survey of 105 topical cannabinoid products purchased online or in stores, THC was detected in 37; four of those were labelled THC-free. This was a selected sample, not a representative estimate for all products or countries. Detection in a cream does not by itself establish how much enters the body or whether it would cause impairment or a positive drug test.14

A separate 2025 analysis of 56 selected US CBD gummy products detected THC in 22 and found that measured CBD differed from the label by more than 10% in 70% of products. THC was not necessarily absent from every label. The researchers also found variation between individual gummies in a subset of products. A certificate of analysis is useful only if it is reliable and relevant to the actual batch. See the driving discussion above and the THC profile; legal and drug-testing rules vary by jurisdiction.15

Evidence is not a general permission slip

A proposed mechanism, a change in a biological marker or a customer testimonial is not the same as a regulator’s finding about a specific product. Ask whether the people, product, comparison and outcome match the claim. A small study that does not find a clear safety signal does not prove safety; an observed association may have other explanations.13

Five questions for a CBD claim

Is the claim about purified CBD, a prescription medicine, a plant extract or an unverified consumer product? Which country, label, use, age group and way of taking it apply? Were the people studied similar to those in the claim? Were liver effects, drowsiness, interactions, contamination and treatment stopping measured for long enough? Does the source show a clinical benefit, or only a laboratory finding or early association?45

Development & regulatory context

These selected records are not a complete global legal-status list. Approval depends on the product, use, population and country; check each record’s date.