The strongest trial of cannabis for chronic insomnia found that people rated their insomnia as clearly less severe while taking a THC-based oil. On the same nights, a full sleep-laboratory recording found almost no change in how they actually slept. This part explains both results and what THC does to dream sleep. Part 2 covers the main reason for caution: what happens to sleep when regular use stops.
This is general information, not medical advice. It is not a recommendation to use cannabis for sleep. Chronic insomnia is usually a symptom of something else. Common causes are pain, anxiety, depression, sleep apnoea (breathing that repeatedly stops during sleep), a medication or a sleep schedule. If only the symptom is treated, the real cause can go unfound for years, even when it has a treatment. In Thailand, cannabis flower is a controlled herb. It may be sold only against a PT 33 prescription (ภ.ท.33) from an authorised practitioner, and each prescription covers no more than 30 days.
1. How a Night of Sleep Is Built
Sleep runs in cycles of roughly 90 minutes. Each cycle starts with lighter stages of non-REM sleep, called N1 and N2. It then moves into deep sleep, also called slow-wave sleep or N3. Most deep sleep comes in the first half of the night.
Each cycle ends with REM sleep, short for “rapid eye movement” sleep. REM makes up roughly a fifth to a quarter of adult sleep. The REM periods get longer towards morning. REM is the stage linked to vivid dreaming, and it is thought to help the brain process emotional memories.
Because sleep has this structure, falling asleep faster does not make something a good sleep aid. A substance can shorten the time it takes you to fall asleep and still change how much deep sleep or REM you get. Section 5 shows that THC does change REM.
What you can do: if you and your prescriber are considering any sleep aid, ask two questions. The first is how much faster it helps you fall asleep. The second is what it does to deep sleep and REM.
2. The First Controlled Study: Faster Sleep, Groggy Mornings
Two researchers, Cousens and DiMascio, published the earliest controlled study of THC as a sleep medicine in 1973. They gave THC (tetrahydrocannabinol, the main intoxicating compound in cannabis) to people with insomnia. They tested three doses: 10, 20 and 30 mg.
People fell asleep faster. Side effects rose as the dose went up. At 30 mg, the next-day “hangover” (grogginess the morning after) was bad enough that the researchers ruled out that dose as a sleep medicine.
Most popular accounts of cannabis and sleep still rest on this picture: you fall asleep faster, and you feel more groggy the next day as the dose goes up. The effect on REM sleep usually gets only a passing mention. Section 5 explains why it deserves more attention.
What you can do: if a prescriber starts or raises a THC dose for you, expect that you may be groggy the next morning. Do not drive or operate machinery that morning until you know how the new dose affects you.
Translator’s note: in Thailand and much of Southeast Asia, many people ride a motorbike to work early in the morning. The same caution applies to riding as to driving a car.
3. The Best Trial So Far: Walsh and Colleagues, 2021
The strongest trial in this area was run by a research team led by Walsh. It was published in the journal SLEEP in 2021. It tested a cannabinoid medicine called ZTL-101 in adults with chronic insomnia.
How the trial was designed. Each feature of the design has a purpose:
- Randomised: chance, not the researchers, decided which product each person took first. This means the researchers could not choose who got what.
- Placebo-controlled: the product was compared with a placebo, an oil that looks the same but has no active ingredient. This shows how much of any change comes from the product itself.
- Double-blind: neither the participants nor the researchers were meant to know which oil was being taken. This is meant to keep expectations from shaping the results.
- Crossover: every participant took both the active oil and the placebo, for two weeks each. This let each person be compared with themselves.
- Washout: a one-week break with no product came between the two periods. This lets the first product clear from the body before the second one starts.
Who took part. 24 people were randomised. 23 completed the trial and were analysed. Their average age was about 53, and 83% were women.
The product. It was an oil taken sublingually, meaning held under the tongue. Each millilitre held 20 mg THC, 2 mg CBN and 1 mg CBD, plus terpenes (the aroma compounds of the plant). CBN (cannabinol) forms as THC breaks down when cannabis ages. CBD (cannabidiol) is a cannabinoid that does not cause intoxication.
The dose schedule. Participants took 0.5 mL about an hour before bed. From night 4, they could increase to 1.0 mL if a doctor approved. They could reduce the dose at any time. This was the trial protocol. It is not a recommended dose.
How the results were measured. The trial used three tools:
- Insomnia Severity Index: a questionnaire of 7 questions on which patients rate how severe their own insomnia is. The total score runs from 0 to 28. This was the primary outcome, meaning the main result the trial was designed to test.
- Actigraphy: a wrist-worn movement monitor, worn at home. It estimates sleep from how still you are.
- Polysomnography: a full overnight recording in a sleep laboratory. It records the brain’s electrical activity and identifies the sleep stages described in section 1.
The table also gives two statistics. The p value shows how likely a difference this large would be if the product had no real effect. By convention, a value below 0.05 counts as statistically significant. The d value (effect size) shows how big the difference is compared with normal variation between people. By convention, 0.8 or more counts as a large effect.
The results.
| Measure | Result compared with placebo |
|---|---|
| Insomnia Severity Index (patient’s own rating) | −5.07 points (p = 0.0001, d = 0.94) |
| Total sleep time (wrist monitor) | +33.4 minutes (p < 0.001) |
| Time awake after first falling asleep (wrist monitor) | −10.2 minutes (p = 0.002) |
| Sleep efficiency, the share of time in bed spent asleep (wrist monitor) | +2.9% (p ≤ 0.005) |
| Polysomnography (sleep stages and sleep structure) | No significant difference |
Side effects. All side effects were mild. On the active oil, 33% of participants had a dry mouth, 25% had dizziness and 17% had a headache. Almost all of these effects (97.5%) went away overnight or soon after waking. There were no serious adverse events.
Compared with most cannabis sleep research, this is a positive trial. The main outcome improved by a large amount, and the wrist monitor supported it with objective data.
What you can do: when a product page or a forum post cites this trial, check whether it also gives the sleep-laboratory result, the 23 participants and the two-week length. If any of these is missing, the post is showing you only part of the trial.
4. Two Weaknesses Inside the Trial
The sleep laboratory did not see what the questionnaire and wrist monitor showed. The wrist monitor recorded clear improvements. Polysomnography found no significant difference between the active oil and the placebo. This is possible because the two tools measure different things. The wrist monitor counts stillness as sleep, while polysomnography reads brain activity. A drug that makes you lie still has not necessarily changed your sleep.
Other cannabinoid trials show the same pattern. What patients report improves, but the objective measure does not improve with it. This does not mean the patients are wrong. It means the trial supports a narrower claim: people rated their insomnia as less severe and lay still for longer. It does not yet show that the sleep recorded from the brain changed.
Participants could tell which oil they were taking. The authors say so directly. The active oil had effects people could feel, so the blinding could not be kept. While on the active oil, all 21 participants who answered the question correctly guessed that they were taking it. Two did not answer.
This matters because the main outcome was the patients’ own rating. A 2022 meta-analysis (a study that pools the results of many trials) by Gedin and colleagues, published in JAMA Network Open, looked at cannabis trials for pain. It found that placebo alone produced a moderate-to-large drop in pain scores. In other words, expecting a treatment to work can move a self-rated score a long way by itself.
In the Walsh trial, people knew when they were taking the real oil. So the active period carried the full effect of expectation, and the placebo period carried less of it. Part of the 5-point gap may therefore come from expectation rather than from the drug. The Gedin study was about pain, not sleep. Both outcomes are rated by the patients themselves, though, so the same risk applies.
The trial was small and short. 23 people over two weeks cannot show what happens over months. The honest summary is that this is a promising signal that needs a bigger and longer trial. It is not a settled result.
What you can do: call this evidence “promising but not settled” when you discuss it with your prescriber or on the forum. Ask your prescriber whether a larger or longer trial has been published since 2021.
5. What THC Does to Dream Sleep (REM)
Studies generally report that THC reduces REM sleep. When THC is stopped, REM tends to rebound: for a while, there is more of it than usual. The long-term significance of this is not well established. A 2017 literature review by Babson, Sottile and Morabito found that the effect seems to depend on dose and on tolerance. It also found that the human data are partly mixed.
THC is not the only substance that does this. Some conventional sleeping pills (hypnotics) and antidepressants also reduce REM. This is one of the criticisms made of those medicines too.
This has two consequences. First, a person who dislikes their dreams may find less REM a benefit. One example is a patient with nightmares from post-traumatic stress disorder (PTSD). Second, when use stops, the REM rebound brings unusually vivid and intense dreams for a period afterwards.
The rebound is part of a wider withdrawal pattern after regular use. Part 2 explains why this withdrawal is the most important point for anyone using cannabis for sleep.
What you can do: if you take a prescribed cannabinoid at night, note any change in your dreams in a sleep diary. If you stop, expect vivid dreams for a while, and tell your prescriber about them.
References
- Babson KA, Sottile J, Morabito D. Cannabis, cannabinoids, and sleep: a review of the literature. Curr Psychiatry Rep. 2017. doi:10.1007/s11920-017-0775-9
- Cousens K, DiMascio A. (−)Δ9 THC as an hypnotic: an experimental study of three dose levels. Psychopharmacologia. 1973;33:355–364. doi:10.1007/BF00437513
- Gedin F, et al. Placebo response and media attention in randomized clinical trials assessing cannabis-based therapies for pain: a systematic review and meta-analysis. JAMA Netw Open. 2022;5(11):e2243848. doi:10.1001/jamanetworkopen.2022.43848
- Walsh JH, Maddison KJ, Rankin T, et al. Treating insomnia symptoms with medicinal cannabis: a randomized, crossover trial of the efficacy of a cannabinoid medicine compared with placebo. Sleep. 2021;44(11):zsab149. doi:10.1093/sleep/zsab149
General information, not medical advice. Persistent insomnia deserves a diagnosis, not just a remedy.
Continued in Part 2: Cannabis and Chronic Insomnia: Withdrawal, Cannabinol and Cannabidiol, Cognitive Behavioural Therapy and Planning an Exit (Part 2 of 2)