CBD and sleep: what the research has examined

What has been measured directly

In 2018 a group in Brazil published a randomised, double-blind, placebo-controlled crossover study in Frontiers in Pharmacology, giving healthy volunteers 300 mg of CBD and recording them overnight in a sleep laboratory with polysomnography — the full electrode setup that measures sleep rather than asking about it.

Nothing moved. Total sleep time, sleep efficiency, sleep onset latency, the proportion of each sleep stage and the number of awakenings were all indistinguishable from placebo. The authors were explicit that a single 300 mg dose did not alter the sleep-wake cycle of healthy subjects.

The number everyone quotes

The optimistic figure circulating on this topic — roughly two thirds of people sleeping better — comes from a 2019 case series in The Permanente Journal. Seventy-two adults were given CBD in a psychiatric clinic; sleep scores improved in 66.7% of them in the first month.

Read the second half of that sentence too. The same paper notes the sleep scores fluctuated over time, while the anxiety scores it also tracked held steadier. And the design does most of the work:

• No control group. Nobody received a placebo, and sleep is unusually responsive to expectation.
• Self-reported. Sleep questionnaires and sleep laboratories routinely disagree, which is the whole reason the study above used electrodes.
• People arrive at a clinic at their worst, and worst points tend to be followed by better ones whatever is done.

What is in a sleep product besides CBD

This is the part that changes how the whole category should be read. Look along a shelf of CBD sleep products and count how many are CBD alone.

A product with four active ingredients that seems to work has told you nothing about which of the four did it. That is not an argument against buying one. It is an argument against concluding anything about CBD from having bought one.

What a sleep laboratory measures

The reason the 2018 study carries more weight than the case series is not its size. It is that polysomnography measures things a person cannot report about themselves.

A product can change how a night feels without changing any of those, and it can change one of them without the sleeper noticing. That gap is the whole reason a null result under electrodes sits above a positive result on a questionnaire, and the reason "I slept better" and "my sleep improved" are not the same claim.

What THC does to sleep

CBD is not the cannabinoid with the sleep literature. THC is, and what it shows is more complicated than the reputation.

• It suppresses REM sleep, the stage in which most dreaming happens, which is why heavy users often report they do not dream.
• Tolerance develops to the sedating effect, so the amount that settled someone at the start stops doing it.
• Stopping produces a rebound: vivid dreams and disturbed nights are among the best-documented features of cannabis withdrawal, and they are the reason quitting feels like it made sleep worse.

That matters on a CBD page for one practical reason. Full-spectrum products carry a trace of THC, and the labelling studies keep finding products whose THC content was not what the label said.

What sleep medicine does instead

Chronic insomnia has a first-line treatment, and it is not a pill. Cognitive behavioural therapy for insomnia — sleep restriction, stimulus control, and unpicking the beliefs that keep people awake — is what sleep medicine reaches for first, ahead of medication, on trial evidence that runs to decades.

It is worth naming on a page like this because it is unglamorous, it is not sold in a bottle, and it outperforms everything that is. The practical parts of it are also the least controversial advice in the whole field:

• One wake time, every day. Anchoring the morning does more than anything done at night.
• Bed is for sleep. Out of it when properly awake, back when sleepy.
• Cut the time in bed to the time asleep, then extend it as sleep consolidates.
• Light in the morning, less of it at night.

What CBD does at the doses people take

The dose in the sleep-laboratory study was 300 mg in a single sitting. Most products sold for sleep deliver between 10 and 50 mg a serving. That gap is worth holding in mind whenever a result from a trial is used to sell something: the trial was not testing what is in the jar.

The other reason it matters is that CBD does not behave in a straight line. Where doses have been compared directly, in anxiety, the middle dose did something the highest one did not. Assuming that more is a stronger version of the same thing is the mistake this compound punishes most often.