I’ll admit I expected cannabis to win this one. Of everything people claim the plant does, knocking you out was the claim I’d privately filed under “probably true,” and sleep trouble is one of the reasons people most often give for using it medically, according to a 2022 clinical review. So when the new Cochrane review came out, I went straight to total sleep time. Against placebo, the pooled gain was 7.68 minutes, drawn from six trials with 194 people, and the 95% confidence interval ran from 23.04 minutes less sleep to 38.39 minutes more. I’ve spent longer than eight minutes picking a podcast to fall asleep to.

TOTAL SLEEP VS PLACEBO
7.68 minutespooled estimate across six trials
The 95% confidence interval ran from 23.04 minutes less sleep to 38.39 minutes more. Source: EMJ: Medical cannabis for insomnia shows limited benefits
FROM THE DOCUMENT
effective blinding for a beneficial treatment effect was unable to be achieved
The trial's own authors on why participants knew which weeks they were on the extract. Source: Walsh et al., ZTL-101 randomized crossover trial in chronic insomnia, Sleep (2021)

The review, published in the Cochrane Database of Systematic Reviews in October by researchers at the Cochrane Associate Centre at Hospital Italiano de Buenos Aires and the University of Manitoba, pooled every randomized trial it could find through April 2025. That came to nine trials and 537 adults aged 18 to 76, most with primary insomnia, and 72.66 percent of them women. These weren’t dispensary gummies. Two trials used licensed cannabinoid medicines and seven used patient-specific preparations with measured THC and CBD, taken as tablets, sprays, or oils. Eight compared cannabis with placebo; one compared it with the antidepressant amitriptyline.

The rest of the sleep outcomes looked like total sleep time. Sleep quality across seven trials and 374 people, and daytime functioning, both came back with intervals that crossed zero, while quality of life ranged from a slight decline to a substantial improvement, which is a range wide enough to drive a truck through. An earlier conference version of the review judged most of the underlying trials to be at high overall risk of bias, and the certainty ratings sit at low or very low. Lead author Lucia Varela was careful about what that means: “This does not mean that medical cannabis is ineffective,” she told reporters. The trials can’t tell you either way.

The side effects are harder to wave off. In three crossover trials with 98 participants, the odds of an adverse event on cannabis came out about 11 times higher than on placebo: an odds ratio of 11.25, with a 95% confidence interval of 4.46 to 28.35, rated low certainty. Mostly that meant dry mouth and dizziness, mild stuff, and the sample is small. But that interval sits entirely above no effect, while every sleep interval straddles it. So far the cost shows up more consistently than the benefit.

ODDS OF AN ADVERSE EVENT (OR)
Figure data
ResultValue (OR)Lower boundUpper bound
Cannabis vs placebo11.254.4628.35
Three small crossover trials, low certainty. Mostly dry mouth and dizziness. Source: EMJ: Medical cannabis for insomnia shows limited benefits

The trial where the sleep lab disagreed

The best-known positive result in this field is a 2021 randomized, double-blind crossover trial in Sleep of ZTL-101, a sublingual oil of THC, CBN, and a little CBD. It randomized 24 people with chronic insomnia, 20 of them women, average age 53; 23 finished. Each spent two weeks on the oil and two on placebo, starting at 11.5 mg of total cannabinoids with the option to double to 23 mg. Zelira Therapeutics, the company developing the product, sponsored it, and the paper states the study was performed entirely independent of the sponsor. When the topline results came out in 2020, principal investigator Peter Eastwood called it “the most rigorous clinical trial ever undertaken to assess the therapeutic potential of medicinal cannabis.”

On paper it delivered. Insomnia Severity Index scores dropped about 5 points against placebo. Averaged over each two-week arm, sleep diaries recorded 64.6 extra minutes of sleep and a wrist actigraph, which infers sleep from stillness, logged 33.4 extra minutes. Then, on the 14th night of each arm, participants slept wired to overnight polysomnography, the lab test that reads brain waves directly, and it found no significant difference between oil and placebo in how fast they fell asleep, how long they lay awake, total sleep, sleep efficiency, or sleep stages.

ZTL-101: EXTRA SLEEP VS PLACEBO
64.6minutes
Sleep diary
33.4minutes
Wrist actigraphy
Overnight polysomnography on the 14th night of each arm found no significant difference in total sleep time. Source: Walsh et al., ZTL-101 randomized crossover trial in chronic insomnia, Sleep (2021)

Wait, how do a diary and a brain recording end up an hour apart? Some of it is the instruments. A wrist tracker is objective, but it can score lying still and awake as sleep, and the polysomnography captured one night per arm rather than two weeks. The authors also note they enrolled people on self-reported insomnia criteria, not lab-defined ones. What I couldn’t get past was the blinding. On the extract, all 21 participants who answered the question guessed correctly that they were on the active drug, 17 of them because they felt they were sleeping better. The authors wrote that “effective blinding for a beneficial treatment effect was unable to be achieved.” That doesn’t prove the diary gain was expectation, but when people know which weeks are the drug weeks, a self-reported benefit becomes very hard to separate from hope. Their bodies noticed too: 17 participants logged 36 nonserious adverse events possibly or likely related to the oil, led by dry mouth and dizziness, with two reports of hallucinations, against four events on placebo.

The nights after you stop

None of the Cochrane trials ran longer than 12 weeks, and the reviewers say future studies need longer follow-up. The questions that matter most for a nightly sleep aid live past that horizon, and the evidence there comes from people who aren’t insomnia patients on prescribed oil, so read it as a warning light, not a verdict.

The 2022 review in the Journal of Primary Care & Community Health describes improvements in sleep onset that appear to decline with chronic use. In one study of 128 adults over 50 with chronic pain, frequent use was linked to more trouble falling asleep and waking up. Among adults trying to quit cannabis, 67% to 73% report sleep difficulty, and in one survey of 495 people who had tried to stop, 65% said withdrawal sleep disturbance had contributed to at least one relapse. The chemistry also cuts both ways: in a tiny crossover study of eight healthy young adults, 15 mg of THC was sedating while 15 mg of CBD increased wake activity during sleep. The ratio in the bottle matters, and a product sold for sleep can contain the molecule that keeps you up.


Not everyone reads the evidence the way Cochrane does. In September, NORML promoted a Sleep Medicine meta-analysis by investigators in Pakistan and Egypt that pooled 10 randomized trials from 2021 to 2025 with more than 2,100 people and concluded cannabinoids sped up sleep onset, lengthened sleep, and improved its quality, with mild, transient side effects. Its question was broader: people with sleep disorders generally rather than diagnosed insomnia, and melatonin as well as placebo as the comparison. Cochrane narrowed to insomnia and graded how much it trusted each result. Those are different questions, and the narrower one is the one a person lying awake with chronic insomnia is asking.

I understand why people go looking. The Cochrane team’s own background notes that conventional insomnia drugs can bring dependence and cognitive problems. But the first-line treatment the American College of Physicians has recommended since 2016 is cognitive behavioral therapy for insomnia, CBT-I, which the college says likely carries fewer harms than medication and which you can do one-on-one, in a group, by phone, online, or from a self-help book. Nobody sells it in a tincture, which may be why you hear about it less.

If I were lying awake at 3 a.m. tonight, I’d start a CBT-I program before buying anything labeled for sleep, and I wouldn’t pay for a cannabis sleep product until a trial holds its blind and the sleep lab agrees with the diary.

Sources

  1. Varela et al., Medical cannabis for the treatment of insomnia, Cochrane Database of Systematic Reviews (2026)
  2. EMJ: Medical cannabis for insomnia shows limited benefits
  3. Walsh et al., ZTL-101 randomized crossover trial in chronic insomnia, Sleep (2021)
  4. Kolla et al., The Effects of Cannabinoids on Sleep, Journal of Primary Care & Community Health (2022)
  5. Varela et al., AAN 2026 conference abstract
  6. ScienceDaily: Scientists find surprisingly little evidence that medical cannabis helps you sleep
  7. News-Medical: Medical cannabis benefits for insomnia remain uncertain, Cochrane review finds
  8. NeurologyLive: ZLT-101 topline results (2020)
  9. American College of Physicians: CBT-I as initial treatment for chronic insomnia (2016)
  10. NORML: summary of a Sleep Medicine meta-analysis (2026)