Legal CBD in France: THC Threshold and Uses

Readers who want the French-language detail on this framework, including the arrêté and the European case law behind it, can consult Nessnet's CBD coverage as a starting point. The same site tracks how the rules apply to food and veterinary products, which is where most consumer confusion sits.

What is the THC threshold in France?

The 0.3% figure applies to the dry weight of the plant material. It comes from the French arrêté of 22 August 1990, which lists the hemp varieties that may be cultivated and processed. A separate European Union rule, Regulation (EU) 2021/2115 on the common agricultural policy, uses the same 0.3% figure for hemp eligible for area payments. The two instruments are not identical in purpose, but they converge on the number.

Two practical points follow. First, a product can test below 0.3% THC and still be illegal to sell as a food or a medicine, because the threshold governs the plant, not the marketing category. Second, the 0.3% is measured on dry material, so a lab result reported on a fresh or oil basis is not directly comparable. A certificate that does not state the basis is not evidence of compliance.

The Court of Justice of the European Union addressed the wider question in Case C-663/18, known as Kanavape. The Court held that a CBD product lawfully produced in one member state cannot be banned in another solely because it is derived from the whole plant rather than from fibre or seed. That ruling concerns free movement, not French criminal law, and French courts have read it narrowly. The result is a framework that is legal at the European level and unsettled at the national one.

What does the French regulatory framework actually cover?

Three layers operate at once. The first is narcotics law, which sets the 0.3% line and prohibits THC above it. The second is food law, which treats CBD as a novel food. Under Regulation (EU) 2015/2283, a food that was not consumed to a significant degree in the EU before 15 May 1997 needs authorisation before it can be sold. The European Commission has kept CBD extracts in the novel food catalogue, and several applications remain under review. Until an authorisation exists, a shop selling CBD as an ingredient is selling an unauthorised novel food, even if the THC is below the threshold.

The third layer is veterinary and cosmetic law. Cosmetic use is governed by Regulation (EC) 1223/2009, and CBD is not on the prohibited list, though claims are restricted. Veterinary products fall under a different regime again, and the French agency ANSES has published guidance on hemp-based animal feeds. The three layers do not always agree, and a product can satisfy one and fail another.

For a consumer, the practical test is not the THC number alone. It is whether the seller can name the authorisation under which the product is sold. A food needs a novel food authorisation. A cosmetic needs a compliant safety report. A veterinary product needs a marketing authorisation or a recognised exemption. A certificate of analysis showing 0.2% THC answers a different question.

What did trials find for sleep?

Sleep is the most common reason people give for trying CBD, and the trial evidence is thinner than the market suggests. A 2019 review by Shannon and colleagues in The Permanente Journal examined 72 adults at a psychiatric clinic who received CBD for anxiety or sleep. The authors reported that anxiety scores fell in the first month and sleep scores improved in the first month, with some fluctuation afterwards. The study had no control group, no blinding, and no placebo, so the improvements cannot be separated from expectancy or from the natural course of symptoms.

A 2021 systematic review by Bhagavan and colleagues in Sleep Medicine Reviews looked specifically at cannabidiol for insomnia. The authors found few randomised trials, small samples, and inconsistent outcome measures. They concluded that the evidence was insufficient to recommend CBD for insomnia outside research. That is a limitation stated by the reviewers, not a verdict on any individual product.

A 2022 randomised crossover trial by Crippa and colleagues, published in Neuropharmacology, tested CBD in people with anxiety-related sleep disturbance. It found a signal on some sleep measures at 300 mg per day, with a small sample. The dose is far above what most retail products deliver per serving. A reader comparing a 10 mg gummy to a 300 mg trial dose is comparing two different interventions.

What did trials find for pain?

Pain evidence is more developed for THC-containing preparations than for CBD alone, and that distinction matters for a French reader, because THC above 0.3% is not legally available. A 2015 systematic review by Whiting and colleagues in JAMA examined 79 randomised trials of cannabinoids for chronic pain and other conditions. The authors found moderate-quality evidence that cannabinoids reduced chronic pain, and they noted an increased risk of short-term adverse events such as dizziness and sedation. Most included trials used preparations with THC, often in ratios close to 1:1 with CBD.

For CBD without THC, the picture is different. A 2020 review by VanDolah, Bauer, and Mauck in Mayo Clinic Proceedings concluded that the strongest evidence for pure CBD was in rare paediatric epilepsies, and that evidence for chronic pain was limited and mostly preclinical. A 2018 review by Niesink and van Laar in Frontiers in Psychiatry reached a similar position for pain and anxiety.

This is the point where the French threshold and the clinical literature diverge. The preparations with the better pain evidence are the ones French law restricts. The preparations that are legally sold in France are the ones with the weaker pain evidence. A reader who wants both legal compliance and documented effect is asking for something the current evidence base does not supply.

How should a reader weigh a label?

Start with the certificate of analysis. It should name the laboratory, the batch, the date, and the basis of measurement. A result expressed as a percentage without a basis is not usable. Then check the product category. A food, a cosmetic, and a veterinary product each carry different legal requirements, and a seller who cannot say which one applies is not in a position to advise.

Then check the dose against the trials. Most randomised studies used doses between 100 mg and 600 mg per day. Retail products commonly contain 5 mg to 25 mg per serving. The gap is not a reason to take more, because higher doses in trials were associated with more adverse events, including drowsiness, diarrhoea, and liver enzyme changes in some epilepsy studies. It is a reason to treat a retail label as a starting point for a conversation with a clinician, not as a treatment plan.

Finally, separate the legal question from the clinical one. A product can be lawfully sold in France and still lack evidence for the use a buyer has in mind. A product can have trial support and still be unlawful to sell above the THC threshold. The two questions have different answers, and a single number on a label does not resolve either.

What this desk does not claim

Nothing here is medical advice, and no dose is recommended. The trials cited enrolled specific populations under supervision, and their results do not transfer automatically to a reader at home. CBD interacts with some medicines through the cytochrome P450 system, including certain anticoagulants and anticonvulsants, and that interaction is a reason to involve a prescriber. The French framework is also moving: novel food applications, court decisions, and agency guidance have all changed in recent years, and a page written today may be out of date within a season. Verify the current position before acting on any of it.