CBD, dreams and the REM stage: what the research says

The night unfolds in cycles of about 90 minutes alternating light non-REM, deep non-REM and the REM stage. It is during that last stage, marked by rapid eye movements and almost complete muscle atonia, that the longest and most narrative dreams occur. Remembering a dream depends above all on how you wake: waking during or just after a REM phase markedly increases dream recall.

That mechanism explains why an impression of "dreaming more" is not the same thing as dreaming more. A more fragmented night, with micro-awakenings towards morning, produces that impression without the quantity of REM stage having moved at all. For the basics, our article on the circadian rhythm describes how these cycles are organised across a night.

The idea that cannabis suppresses the REM stage is very widespread. A meta-analysis published in 2025 (Velzeboer et al.) qualifies it seriously. It retained eighteen studies, nine of which were usable statistically. Its authors conclude that administering cannabis does not consistently alter the duration of the night, its onset latency, its efficiency or the distribution of stages. The reductions in REM stage come mainly from older studies, at high THC doses and methodologically limited. More recent work, at doses used clinically, gives mixed or null results.

One result, on the other hand, comes out consistently: stopping an active habit is accompanied by a disturbed night, with a reduction in total rest time, a longer time to drop off and a REM-stage rebound.

That point is decisive for interpreting testimonials. Someone replacing cannabis joints with CBD products changes two things at once: they add cannabidiol and they remove THC. The REM-stage rebound attributed to CBD is then most probably the effect of the THC being withdrawn. Our article on THC withdrawal returns to that transition.

The most direct study was published in Frontiers in Pharmacology in 2018 by Linares and colleagues, under a title that leaves no ambiguity: cannabidiol had no acute effect on the wake-and-rest cycle of healthy subjects. Twenty-seven healthy volunteers received, in a crossover double-blind design, 300 mg of cannabidiol or a placebo 30 minutes before an 8-hour polysomnographic recording.

No measured parameter was significantly altered: neither the time taken to drop off, nor the efficiency of the night, nor the distribution of stages, REM stage included. The authors conclude that these results support the idea that CBD does not alter the normal architecture of the night. In their view, that sets it apart from benzodiazepines and serotonergic antidepressants.

A negative study in 27 people does not close the question. It is nevertheless the cleanest controlled data we have on cannabidiol alone.

A bibliographic search on cannabidiol and dream recall returns almost nothing. The only traceable observation goes back to Carlini and Cunha, in the Journal of Clinical Pharmacology in 1981: in people with insomnia, three doses of CBD (40, 80 and 160 mg a day) were associated with a decrease in dream recall, with no adverse effect the following day. That is the opposite of what is read online, and no modern controlled study has sought to reproduce the result.

On the precise question that brings most people here, the literature therefore holds one observation forty-five years old, pointing the opposite way to the dominant account, and nothing else.

A case series published in 2014 by Chagas and colleagues described four patients with Parkinson's disease presenting REM-stage behaviour disorder. In this parasomnia, muscle atonia disappears and the person acts out their dreams. Three received 75 mg of CBD a day, another 300 mg, over six weeks. Episodes became rarer, then returned to their initial frequency once dosing stopped.

What follows is what the online account almost always leaves out. The same Brazilian team ran a randomised double-blind placebo-controlled trial, published in Movement Disorders in 2021 (de Almeida et al.): 33 patients, 75 to 300 mg of CBD, fourteen weeks of follow-up. The result is clear: CBD showed no difference from placebo on the primary endpoints, and the authors conclude that there was no reduction in the manifestations of the disorder.

Four uncontrolled observations on one side, a negative randomised trial on the other: this is a textbook case of what placebo control changes, and a good reason to be wary of case series cited as proof.

Nothing, where the night is concerned. ANSES recalled in its opinion of 28 November 2024 that, pending EFSA's conclusions, the marketing and consumption of foods containing CBD are prohibited. EFSA also published an update of its position in February 2026. It adopts a provisionally safe level of 0.0275 mg per kilogram of body weight per day, that is about 2 mg a day for an adult of 70 kg, and states that safety cannot be established in the under-25s, during pregnancy and breastfeeding, or where concomitant medication is taken. The full picture is in our article on CBD and the night, and the general framework in our tracking of the Novel Food regulation.

A lasting change in how you rest, repeated nightmares or nocturnal motor behaviours call for advice from a healthcare professional: these signs can reveal a specific disorder that can be diagnosed. What we document on our side is the exact content of our batches, whose analyses are published.