Cannabis, CBD, and Insomnia: What Causes Sleeplessness and What the Evidence Supports

A systematic review in CNS Drugs identified only five studies of cannabinoids in diagnosed insomnia disorder, with 219 participants in total, and rated every one of them poor quality. That remains the honest summary of the field’s foundation.

The trials published since have added precision rather than confidence. A randomized trial of 150 mg nightly cannabidiol found no improvement in insomnia severity. A trial of combined THC and cannabidiol suppressed rapid eye movement sleep. A trial of cannabinol missed its primary endpoint while improving several secondary ones.

Insomnia is a symptom with many drivers rather than a single disease, and the driver determines what helps. A product that reduces hyperarousal will do little for pain that wakes someone at three in the morning, and a sedating product will do nothing about untreated sleep apnea except mask it.

Cannabinoid sleep products are sold with more confidence than the trial data support. Patients who try them, see little benefit, and conclude that nothing works for their sleep are frequently the same patients whose actual cause has never been identified.

Insomnia is a presenting complaint rather than a diagnosis in itself. The common drivers include stress and physiologic hyperarousal, anxiety and mood disorders, chronic pain, irregular or shifting sleep schedules, caffeine and alcohol timing, medications that disturb sleep architecture, and primary sleep disorders such as obstructive sleep apnea and restless legs syndrome.

Those categories respond to different interventions. Hyperarousal responds to behavioral treatment and, in some patients, to anxiolysis. Pain-driven awakening responds to analgesia. Sleep apnea responds to airway treatment and is made worse, not better, by anything that deepens sedation without addressing the airway.

Cognitive behavioral therapy for insomnia remains the first-line treatment for chronic insomnia disorder in every major guideline, and it is the intervention with the most durable effect after treatment stops. Any conversation about cannabinoids that skips over that fact is incomplete.

The most directly relevant trial is a randomized, placebo-controlled pilot published in the Journal of Clinical Sleep Medicine in 2024. Thirty adults with moderate to severe insomnia took 150 mg of cannabidiol or placebo sublingually one hour before bed for two weeks, following a single-blind placebo run-in week. Insomnia severity, self-reported sleep-onset latency, sleep efficiency, and wake after sleep onset did not differ between groups.

Two secondary findings ran in the other direction. The cannabidiol group reported greater well-being throughout the trial, and showed better objective sleep efficiency measured by actigraphy at the two-week mark. The authors read the pattern as more psychological than hypnotic, and with fifteen participants per arm that reading should be held loosely.

A companion analysis of the same cohort, published in Psychopharmacology, tested next-day neurocognitive performance across attention, executive function, reasoning, processing speed, working memory, and episodic memory. Performance was unaffected. Participants on cannabidiol reported more calmness, clear-headedness, and coordination, and more dry mouth. For a sleep aid, an absent next-day cognitive penalty is a genuinely useful finding even when the sleep effect is absent too.

A 2025 pilot randomized trial in the Journal of Sleep Research gave 20 adults with DSM-5 diagnosed insomnia disorder a single oral dose of 10 mg THC with 200 mg cannabidiol, with sleep measured by 256-channel high-density electroencephalography.

Total sleep time fell by about 24.5 minutes. Wake after sleep onset did not change. Time in rapid eye movement sleep fell by 33.9 minutes, a large effect, and latency to REM sleep increased by 65.6 minutes. Next-day objective alertness was unchanged, while self-reported sleepiness rose slightly.

REM suppression is the finding patients most need explained. REM sleep participates in memory consolidation and emotional processing, so suppressing it is a trade rather than a benefit. It also sets up rebound: people who use THC nightly for months and then stop frequently experience a period of vivid dreaming and fragmented sleep, which is often misread as proof that they needed the THC.

An older single ascending-dose trial in nine participants, included in the CNS Drugs review, found that THC at 10, 20, and 30 mg shortened sleep-onset latency compared with placebo by roughly 43 to 62 minutes. Nine participants is not a foundation for clinical guidance, and falling asleep faster is not the same as sleeping better.

Cannabinol has been marketed as a sleep cannabinoid for years on essentially no human evidence. That changed in 2026, when the CUPID trial was published in the Journal of Sleep Research.

Twenty adults with physician-diagnosed insomnia disorder and an Insomnia Severity Index of 15 or higher received single oral doses of 30 mg cannabinol, 300 mg cannabinol, and matched placebo on three separate nights, each separated by a two-week washout, with overnight polysomnography. The primary outcome was wake after sleep onset.

Cannabinol did not significantly change wake after sleep onset at either dose. The 300 mg dose did increase stage N2 sleep, improve subjective sleep quality, shorten sleep-onset latency, and reduce electroencephalographic arousal indices. There were 247 mild to moderate adverse events across the three arms.

That is a mixed result and should be described as one. A trial that misses its primary endpoint while moving several secondary measures generates a hypothesis worth testing in a larger and longer study. It does not establish that cannabinol treats insomnia, and it certainly does not validate the doses present in commercial sleep gummies, which are a small fraction of 300 mg.

The most convincing cannabinoid sleep result comes from a population defined by something other than insomnia alone. In a randomized, double-blind crossover trial in Anesthesia and Analgesia, 29 patients with fibromyalgia and chronic insomnia received nabilone, a synthetic THC analogue, at 0.5 to 1.0 mg at bedtime and amitriptyline at 10 to 20 mg at bedtime for two weeks each.

Nabilone outperformed amitriptyline on the Insomnia Severity Index by 3.2 points, with a confidence interval from 1.2 to 5.3. It was marginally better on restfulness and no better on wakefulness. It produced no effect on pain, mood, or quality of life, and adverse events, mainly dizziness, nausea, and dry mouth, were more frequent than with amitriptyline.

The pattern across this literature is consistent enough to be worth naming. Where a cannabinoid helps sleep, it tends to do so in patients whose sleep is being disrupted by something else the cannabinoid affects, and it tends to be a pharmaceutical-grade compound at a known dose rather than a retail product.

This is a weak evidence base being asked to carry a large amount of consumer demand. The trials that exist are small, most enroll twenty to thirty participants, most run for two weeks or a single night, and the systematic review that pooled the earlier work rated every included study poor quality.

Within that weakness there are gradations worth respecting. Polysomnography and high-density electroencephalography are strong measurement tools, so the sleep architecture findings from the THC and cannabinol trials are more trustworthy than any of the self-reported outcomes. What those tools measured, however, was mostly the absence of the expected benefit.

Sample sizes of fifteen to twenty per arm cannot exclude clinically meaningful effects, which cuts both ways. The cannabidiol trial’s null result on insomnia severity is compatible with a real but modest benefit that the study was too small to detect, and its positive secondary findings are equally compatible with chance.

Single-night crossover designs, used in both the THC and cannabinol trials, tell you about acute pharmacology and nothing about what happens on night sixty. Insomnia is a chronic condition and tolerance to the sedative effects of cannabinoids is well documented, so acute findings should not be extrapolated to nightly use.

Almost none of this work tests the products patients actually buy. Retail cannabinol gummies contain a few milligrams, not 300. Full-spectrum tinctures contain variable amounts of several cannabinoids and terpenes. Trial results obtained with a defined single compound at a defined dose do not transfer to an undefined mixture.

No trial has shown that cannabidiol improves insomnia severity in a diagnosed insomnia population. The best-designed test of that question returned a null result on its primary measures.

No trial has established a dose, a formulation, a timing, or a duration for any cannabinoid in insomnia. There is no evidence-based cannabinoid sleep protocol, and anyone presenting one is extrapolating.

Nothing here addresses long-term use. The consequences of nightly REM suppression over months or years, the trajectory of tolerance, and the difficulty of discontinuation after prolonged use are unstudied in controlled trials, despite being the most common real-world pattern.

The gap between the commercial confidence around cannabinoid sleep products and the state of the evidence is one of the widest in this field. Sleep is an enormous consumer market, subjective improvement is easy to feel and hard to verify, and insomnia has a substantial placebo response, which together create ideal conditions for products that outrun their data.

The cannabinol result is a useful case study in how this works. A cannabinoid acquired a reputation as a sedative through repetition rather than research, an entire product category was built on it, and the first rigorous human trial arrived years later and missed its primary endpoint. The trial is genuinely interesting science. It is not a vindication of the products that preceded it.

I tell patients the same thing I would want told to me, which is that the evidence for cannabis and sleep is thinner than they have been led to believe. That disappoints people, and I would rather disappoint someone early than have them spend a year and several hundred dollars discovering it themselves.

What I have seen work is narrower than what is marketed. Patients whose sleep is wrecked by anxiety or by pain sometimes sleep better when the anxiety or pain improves, and a cannabinoid can be part of that. Patients whose insomnia is its own problem, with no clear driver, rarely get durable benefit from cannabis alone, and some of them end up dependent on it for sleep in a way that is harder to unwind than the original problem.

The REM finding is the one I spend the most time on in the room. Someone using THC nightly for a year who tries to stop will have a rough few weeks of vivid dreams and broken sleep, and they will interpret that as proof they need it. It is worth knowing that in advance.

And before any of this, I want to know whether the person has been evaluated for sleep apnea. Sedating an untreated airway problem is the one genuinely dangerous mistake available in this conversation.

Cannabis and cannabidiol are not established treatments for insomnia disorder, and they should not be offered as first-line therapy. Cognitive behavioral therapy for insomnia is, and the cause of the sleeplessness should be identified before any medication is chosen. Where cannabinoids have a role, it is usually in patients whose sleep is disrupted by pain or anxiety that the cannabinoid addresses, and that decision belongs in a clinical conversation that accounts for other medications, sleep apnea risk, and what has already been tried.

The finding to carry forward is that this literature is small, short, and mostly null on its primary endpoints, while being more informative about sleep architecture than about symptom relief. The finding not to carry forward is any claim that a specific cannabinoid, dose, or product has been shown to treat insomnia. That claim does not exist in the trial record.