CBD: the ingredient, the labels and the questions

CBD turns up as an oil, a gummy, a capsule, a face cream and, in one case, a prescription medicine kept under medical supervision. The word on the box is the same. The thing inside, and the evidence behind it, is not. So what is CBD, which of its reputations have trials behind them, and how do you read the label without squinting?

In a nutshell: CBD (cannabidiol) is a non-intoxicating compound from the cannabis plant. It has one well-evidenced use, as a prescription medicine for certain rare childhood epilepsies. For anxiety, sleep and pain, the human evidence is much thinner and mixed. On the legal side, EFSA's February 2026 statement says oral CBD products remain unauthorised in the EU, and the Commission said in February 2023 that no CBD or other cannabinoid had been authorised as a food; we found no later authorisation. That is why the label and the legal context are worth a careful read.

Meet the molecule: CBD, THC and the rest of the family

Cannabis contains roughly 125 cannabinoids, according to the introduction of a 2025 rat study from the University of Sydney's Lambert Initiative. Two are famous. THC is the main intoxicating one. CBD is the main non-intoxicating one. Hemp is the cannabis grown for industrial uses, fibre, seed and CBD extract, and is generally low in THC.

"Non-intoxicating" is a statement about CBD's effect on the mind, not about its safety. The Court of Justice of the EU judged in 2020 (case C-663/18) that CBD should not be treated as a narcotic drug under the 1961 UN convention because it does not have a psychotropic effect. The European Commission relayed that reasoning in a 2023 statement, adding that CBD can in principle be a food, but only if it clears the novel food rules covered below.

The other names you will meet on labels are minor cannabinoids with their own stories: CBN (a breakdown product of THC) and CBG (a precursor of several others). Evidence for one tells you nothing about another, so they get their own guide.

One practical point: hemp extracts are rarely pure CBD. A "full-spectrum" oil also contains other cannabinoids, including trace THC. That matters for the label, for drug tests and for the law, and we come back to all three.

Does it work? A ladder from strongest evidence to thinnest

Think of the evidence as a ladder. The rungs near the top have randomised trials and a regulator's approval. The rungs near the bottom have hopeful studies, small samples and a lot of marketing.

Rung one: prescription CBD and rare epilepsies

The European Medicines Agency authorised Epidyolex on 19 September 2019. It is a liquid containing 100 mg of CBD per ml, indicated as add-on therapy for seizures in Lennox-Gastaut syndrome and Dravet syndrome (together with clobazam) in people aged two and over, and also for seizures in tuberous sclerosis complex. The EMA lists sleepiness, decreased appetite, diarrhoea, fever, tiredness and vomiting among the most common side effects, and raised liver enzymes as the most common reason people stopped treatment.

The trials behind it are real and sizeable for such rare conditions:

• In a Dravet syndrome trial of 120 children and young adults, 20 mg of CBD per kilogram of body weight per day, added to existing treatment, cut the median monthly convulsive seizures from 12.4 to 5.9. The placebo group went from 14.9 to 14.1.
• In a Lennox-Gastaut trial of 225 people, drop seizures fell by a median of 41.9% on 20 mg/kg, 37.2% on 10 mg/kg and 17.2% on placebo. The listed funders include GW Research, the company behind the medicine.
• A second Lennox-Gastaut trial of 171 people, run in sites including the Netherlands, found median drop seizures down 43.9% with CBD versus 21.8% with placebo.
• A systematic review of 36 studies concluded that pharmaceutical-grade CBD as an add-on in drug-resistant childhood-onset epilepsy may reduce seizure frequency, while rating the certainty of the evidence as low. It also found more adverse events, including serious ones, than placebo.

Two things stand out when you compare this with a shop bottle. These doses were around 20 mg per kilogram, so roughly 1,400 mg a day for a 70 kg adult, given as a standardised medicine under supervision. And the participants had severe, treatment-resistant epilepsy. The result is a finding about that medicine in that population, not a general endorsement of every CBD product.

Rung two: anxiety

Here the picture is promising but unsettled. A 2024 meta-analysis of eight trials with 316 participants found a sizeable average effect of CBD on anxiety, but the confidence interval was wide (Hedges' g of −0.92, 95% CI −1.80 to −0.04), and the authors themselves urged caution because of the small sample. Another 2024 review of 11 randomised trials found that results across anxiety disorders and doses were often contradictory, even though CBD seemed to reduce anxiety with few side effects compared with placebo.

Individual trials show why. In 63 people with high trait worry, 300 mg of CBD did not change worry after one dose or two weeks, though it reduced anxiety symptoms over the two weeks. Canopy Growth, a cannabis company, funded that trial. In 32 students facing a stats exam, single doses of 150, 300 or 600 mg did nothing for test anxiety.

The honest summary: some signal, small studies, varying doses (often 300 mg or more, well above many retail servings), and no agreed dose or population where CBD clearly works.

Rung three: sleep

A 2023 systematic review of 34 studies on CBD and insomnia found that every study reported improvement in at least some participants. But only two of the 34 studies focused on people with insomnia, one of them a case report, and many used unvalidated questionnaires. The authors said more research was needed before firm conclusions.

Two newer trials illustrate the mixed picture. In a pilot of 30 adults with insomnia, 150 mg of CBD nightly for two weeks was similar to placebo on most sleep measures, with better well-being scores and, on wrist actigraphy, higher sleep efficiency after two weeks. In a crossover study of 125 people, 300 mg of CBD plus a terpene blend nudged the share of the night spent in deep and REM sleep up by about 1.3 percentage points, as measured by a wrist-worn tracker. That is a measurable shift in a combination product, not proof that CBD alone helps people fall asleep.

Rung four: pain

Pain is where the headlines and the data part ways most. The Cochrane review of cannabis-based medicines for chronic neuropathic pain (updated in 2026, 21 studies, 2,187 participants) looked at THC-dominant, balanced THC and CBD, and CBD-dominant medicines. For CBD-dominant products (five studies, 208 participants), it found no clear evidence of an effect on the number of people getting at least 50% pain relief, with very low certainty. A broader BMJ review of 32 trials found small to very small improvements from medical cannabis or cannabinoids as a group. Because it pooled products of different compositions, it cannot be read as a result for CBD on its own.

What to take from the ladder

Strong evidence for a prescription medicine does not transfer to a shop bottle. For anxiety, sleep and pain, CBD is an interesting ingredient with unproven benefits in people. If you hear "studies show CBD helps with X", ask which product, which dose, which population and who paid.

Reading a CBD label without squinting

CBD labels cheerfully mix three units, so let's untangle them.

• Percentage on an oil. A "5%" oil usually means 5% CBD by weight or volume in the carrier oil. In a 10 ml bottle that works out at about 500 mg of CBD in total, or around 50 mg per ml. A worked example with neutral figures: if a label gives 250 drops per 10 ml bottle, 500 mg means about 2 mg per drop, and 1,000 mg would mean about 4 mg per drop.
• Per drop. This depends on the dropper, so treat it as an approximation. If a bottle's label gives 25 drops per ml and 50 mg per ml, that is about 2 mg per drop.
• Per serving. For capsules and gummies the number to read is mg per piece, then multiply by the pieces in a serving and the pieces in the pack. Thirty gummies at 25 mg each is 750 mg in the pack, which is not the same as 25 mg per day of anything.

A common slip is to compare a bottle's total milligrams with another product's per-serving milligrams. Divide everything to the same unit, per serving, before you compare.

Spectrum terms

"Full-spectrum", "broad-spectrum" and "isolate" are marketing categories for how much of the plant's other compounds remain. Broadly: full-spectrum products keep the other cannabinoids, including trace THC; broad-spectrum products aim to remove THC; isolates aim for CBD alone. A 2024 analysis of 202 CBD products bought in the US found that 26% did not match the product type on the packaging, and 74% deviated from the stated CBD content by at least 10%. The study was funded by Jazz Pharmaceuticals, which sells a prescription CBD medicine, and sampled the US market, so it is a warning about checking, not a verdict on European products.

The certificate of analysis (COA)

A good COA is the best tool you have. Look for:

1. A batch or lot number that matches your bottle.
2. The name of the lab, preferably independent, and the date.
3. The cannabinoid profile with CBD per ml or per piece, and THC reported, not just "pass".
4. Tests for contaminants (heavy metals, pesticides, solvents), not only cannabinoids.

If the COA covers a different batch, or only shows CBD, you have less information than the shop page suggests. Our guide to reading supplement labels covers the general habits that apply to every bottle.

What one brand's range says about itself

This section describes what Cibdol's public pages said on 8 October 2026, as supplier information; we haven't tested the products. It describes the supplier's pages, not how any product is classified in a given country.

Its site lists CBD oils in six strengths from 5% to 40%, capsules, gummies, CBG and CBN oils, and sleep products. Its oil page pairs the 5% oil with 500 mg per 10 ml bottle and about 2 mg per drop, which matches the label arithmetic above. It describes its main oil as full-spectrum, says each batch is analysed by an independent laboratory, and offers a batch lookup for the certificate. Those are the supplier's claims, and the batch certificate is the document that lets you check them. Health claims about any product still need the kind of human evidence described above, which a supplier page cannot supply.

Is it allowed? CBD in the EU and the Netherlands (read on 8 October 2026)

Laws change and differ between countries; check current rules before buying or travelling.

EU: novel food status

Under the EU's novel food rules, a food without a significant history of consumption in the EU before 15 May 1997 needs authorisation before it can be sold. The Commission's statement of 27 February 2023 said that this history had not been shown for CBD or other cannabinoids from the cannabis plant, so they count as novel foods, and that no CBD or other cannabinoid had been authorised. The statement said the Commission had received over 190 applications, of which 20 had been judged valid at that point. We didn't search the Union list of authorised novel foods ourselves, so the points below are attributed to what EFSA and the Commission said; the Commission's Novel Food Status Catalogue is the place to check a specific product.

Since then:

• On 9 February 2026, EFSA published an updated safety statement which says CBD products intended for oral consumption remain unauthorised under the novel food regulation.
• The Commission's page of procedure-termination decisions lists many CBD-related applications closed without updating the list of authorised novel foods, including several in 2026.
• In September 2026 EFSA published a positive opinion on one CBD novel food: a chemically synthesised CBD at 98% purity or more, in MCT oil, at a maximum of 2 mg a day for adults over 25, with safety not established for people taking medication. In the same month, EFSA could not establish the safety of a CBD isolate at up to 60 mg a day and a hemp extract at up to 30 mg a day, because the applicant in each case did not supply data EFSA asked for. EFSA does not authorise anything itself; the Commission decides, and we found no authorisation decision as of 8 October 2026.

The 2 mg figure, checked

You may see "EFSA says 2 mg a day" quoted online. That is accurate, but it needs its conditions. EFSA's 2026 statement derived a provisional safe intake of 0.0275 mg per kg of body weight per day, approximately 2 mg a day for a 70 kg adult, using an uncertainty factor of 400. It applies only to food supplement formulations with CBD purity of at least 98%, no nanoparticles, a production process considered safe and genotoxicity ruled out. EFSA's wording is that the figure applies "solely" to such supplements. It said safety could not be established for people under 25, pregnant or lactating women and people taking medicines. It describes the number as provisional, to be revisited as new data arrive.

It is a risk-assessment number for a specific kind of ingredient, not a recommended dose and not a verdict on any particular retail product. It does show how far apart the regulator's cautious figure and the amounts in many retail labels and trials sit.

The Netherlands

The Opium Act (Opiumwet) text, in force from 17 September 2026 and read on 8 October 2026, lists tetrahydrocannabinol on List I and also names "hennepolie" (a concentrate obtained by extracting hemp or hashish) on List I, while "hennep" (any part of the cannabis plant from which the resin has not been extracted, except the seeds) and hashish are on List II. CBD itself is not named on either list.

We found no official Dutch THC limit for CBD products. DRUGSinfo (undated) mentions 0.2%; undated law-firm commentary mentions 0.05%. We state no limit of our own. The statute text doesn't say whether a given hemp-extract oil falls under "hennepolie", and we found no official statement on it. THC content, novel food status and the claims on the pack are three separate questions, and a product can pass one and still raise another.

Not intoxicating, not trivial: safety essentials

Not intoxicating doesn't mean harmless. What the regulators and researchers actually flag:

• Liver. EFSA's 2026 statement lists liver effects among its main concerns, noting consistent liver findings in animal studies and hepatotoxic potential in human trials, especially alongside other medicines. In the Dravet syndrome data EFSA reviewed, 69 patients (22%) on the prescription medicine had liver enzymes above three times the upper normal limit, and 84% of those were also taking valproic acid. The EMA's information for Epidyolex says it must not be used in people with certain combined liver test results.
• Other data gaps. EFSA also names reproductive and developmental toxicity, endocrine effects and the nervous system. Pregnant and breastfeeding women are singled out because of the reproductive and placental findings; under-25s because EFSA notes that neuronal development continues to about the age of 25 and the data on it are inadequate.
• Drowsiness. Sleepiness was among the most common effects in the epilepsy trials, and the FDA's consumer page on CBD lists changes in alertness, most commonly drowsiness. Keep that in mind before driving or doing anything that needs concentration.
• Medicines. The FDA states that CBD can affect how other drugs work. If you take any prescription medicine, ask a pharmacist about your exact combination. Our overview of supplement and drug interactions explains the general logic.
• Drug tests. A 2026 systematic review of 12 studies on CBD and hemp oil and drug test results found that five of eight human studies reported no THC-positive result above the cut-offs, while noting trace levels near detection limits can occur, since some products contain trace THC through manufacturing variation or contamination. If a workplace or sport tests you, that is worth asking about first.

Where that leaves a curious shopper

CBD is a real compound with a real medicine behind it and a lot of marketing in front of it. A practical way to approach it:

• Ask what the product claims, and whether any human trial of this ingredient, dose and population backs that claim.
• Read the label in per-serving units and ask for the batch COA.
• Remember that EFSA's February 2026 statement and the Commission's 27 February 2023 statement both describe oral CBD as unauthorised under the novel food rules, and that Dutch practice is a patchwork of Opium Act text, tolerance policy and enforcement priorities.
• If you are under 25, pregnant, breastfeeding or take medicines, EFSA says safety cannot be established; talk to a pharmacist or doctor first.

If the trigger for your CBD curiosity is sleep, the habits in our sleep hygiene guide have better evidence than any bottle.

Where to read next

On the legal side:

• Is CBD Legal in Europe? EU Law and the Dutch Position, for how EU novel food status, EFSA's 2 mg figure, THC numbers and the Dutch position fit together.

Written by Smart Supplements with AI assistance from the sources linked in the text. It has not been reviewed by a clinician.

Last updated: 12 October 2026

This article is for informational purposes only and is not intended as medical advice. Always consult a healthcare professional before starting any new supplement, especially if you take prescription medication.