THC and sleep: what the trials found

What THC does to a night of sleep

The measurable effects are narrow and they are consistent across the older laboratory work summarised by Babson and colleagues in Current Psychiatry Reports: sleep latency, the time between lights out and sleep, gets shorter. Slow-wave sleep increases somewhat. REM sleep falls.

What does not appear is the thing people actually want, which is more total sleep of better quality night after night. The laboratory studies are short, most of them run on people who already use cannabis, and the effects that show up on the first few nights are not the effects that show up on the twentieth.

The REM question

REM suppression is the most reliable finding on this page and it cuts both ways. People who do not want to dream — and that is a real reason some users give — get what they came for. Everyone else is trading away the stage of sleep most closely tied to memory consolidation and emotional processing, without being told they are trading anything.

It also sets up the rebound described further down. Suppressed REM does not stay suppressed once the THC stops; it comes back louder.

Tolerance arrives within days

This is the part the product copy never carries. The sleep-latency effect and the REM suppression both fade with repeated nightly use, and the fading is fast — days to a couple of weeks, not months. What is left is a habit that no longer does the thing it was adopted for.

Anyone who has used THC nightly for a year and believes it is still getting them to sleep is usually describing the absence of withdrawal rather than an active effect.

Coming off it makes sleep worse

The sleep apnoea trials

The strongest evidence on this page is not about insomnia at all. The PACE trial, published in Sleep in 2018, randomised 73 people with obstructive sleep apnoea to placebo, 2.5 mg or 10 mg of dronabinol — synthetic THC — daily for six weeks.

The 10 mg dose reduced the apnoea-hypopnoea index dose-dependently and improved self-reported sleepiness against placebo. That is a real result from a real randomised trial, and it is the reason this topic keeps coming up.

Why sleep doctors said no

Within months of that trial, the American Academy of Sleep Medicine published a position statement in the Journal of Clinical Sleep Medicine. Its conclusion was that medical cannabis and its synthetic extracts should not be used for obstructive sleep apnoea, citing unreliable delivery methods and insufficient evidence on effectiveness, tolerability and safety.

It went further and recommended that sleep apnoea be excluded from state medical cannabis programmes. One positive phase II trial of a measured pharmaceutical dose is not a reason to inhale an unmeasured one, and the specialty that owns the condition said so in print.

What the systematic reviews found

Suraev and colleagues reviewed the preclinical and clinical literature on cannabinoids across sleep disorders in Sleep Medicine Reviews and reached the conclusion this whole field keeps reaching: promising enough to keep studying, nowhere near settled enough to act on. Almost every trial they could include used a pharmaceutical cannabinoid at a known dose.

THC and CBD are not the same here

They are frequently sold as interchangeable sleep aids and they behave differently. THC is sedating at low doses and activating at higher ones; CBD is not sedating in the same way and the evidence for it in sleep is thinner still. Our page on what the research has examined on CBD and sleep covers that side.

Dose is not a straight line

The dose-response work that matters here was done on stress rather than sleep, and it is the most useful single finding in this section. Childs and colleagues gave 7.5 mg or 12.5 mg of oral THC before a standardised stress task: the lower dose reduced subjective distress, the higher dose increased negative mood and made performance worse.

A curve that turns over like that explains the commonest complaint about THC at bedtime, which is that a bit more did the opposite of a bit.

Onset and duration by method

Method	Onset	Peak	Roughly how long
Inhaled	Minutes	10–30 min	2–4 hours
Sublingual	15–45 min	1–2 hours	4–6 hours
Edible or capsule	30 min–2 hours	2–4 hours	6–8 hours

Those windows are why the delivery method matters more for sleep than for anything else. An inhaled dose taken at eleven is largely gone by two; an edible taken at eleven is still working at six, which is where the morning grogginess comes from.

Edibles and the 3am problem

The commonest pattern people describe is falling asleep quickly and waking in the small hours, and the arithmetic above is usually the explanation. An inhaled dose has worn off by then and whatever kept the person awake in the first place is still there, unaddressed.

What a sleep label cannot tell you

• “Indica” is not a sedation rating. The label describes plant lineage, not effect, and two products carrying it can behave nothing alike.
• Added cannabinoids such as CBN are marketed for sleep on evidence that is thinner than the THC evidence, which is itself thin.
• Melatonin in the same gummy does real work that then gets attributed to the cannabis.

Alcohol, and the same mistake

Alcohol also shortens sleep latency, also suppresses REM and also fragments the second half of the night. It is the closest familiar analogue to what THC does at bedtime, and nobody now regards a nightcap as a sleep aid. The reasoning that retired the nightcap applies here.

Age, and why older users differ

Older adults report more next-day impairment and more dizziness at the same dose, and they are more likely to be taking something else that sedates. The harms columns in the pain meta-analysis below are drawn from mixed-age trial populations and skew accordingly.

Driving the morning after

When the problem is not sleep

Insomnia that has lasted three months or more, most nights of the week, is a condition with a first-line approach that is not a drug at all: cognitive behavioural therapy for insomnia, which outperforms sedatives in head-to-head trials and keeps working after it stops. Untreated sleep apnoea, restless legs, pain and anxiety all present as insomnia and all have their own answers.

What to raise with a doctor

• Whether you snore or stop breathing in your sleep, which points at apnoea.
• Everything else you take at night, sedating or not.
• How long this has been going on, because three months is the line that changes the answer.
• That you are using or considering THC, which interacts with more than people expect.

What this page will not tell you

How many milligrams to take. There is no established dose of THC for sleep, the one trial with a dose worth naming used a pharmaceutical capsule for a different condition, and the specialty body that reviewed it recommended against the use anyway. Anyone publishing a milligram figure for insomnia is ahead of the evidence.

Where to read the evidence

The sources below are the reviews and trials this page is built on, and every one of them is readable in full or in abstract without a subscription. If you take one, take the position statement: it is short, it is specific, and it is the clearest statement anywhere of how far the evidence does and does not reach. For what a dose does generally, see how much THC is in a typical joint.