Cannabis Does Not Consistently Change Sleep Architecture, Review Finds, But Withdrawal Does

A systematic review and meta-analysis of 18 polysomnographic studies reveals highly mixed evidence on whether cannabis alters objective sleep parameters, while consistently demonstrating that withdrawal from regular cannabis use disrupts sleep architecture, including reduced total sleep time, prolonged sleep onset latency, and REM rebound.

Why This Matters

Cannabis is now one of the most commonly self-prescribed substances for sleep complaints worldwide, driven largely by subjective user reports and a growing commercial wellness market. Yet the gap between what patients believe cannabis does for their sleep and what objective brain-wave monitoring actually shows has remained poorly characterized. This review arrives at a critical moment: as cannabis access expands through legalization and medical programs, clinicians face mounting pressure to provide evidence-based guidance on its sleep effects without a clear evidence base to draw from.

Clinical Summary

Sleep complaints are among the top reasons patients cite for cannabis use, and the endocannabinoid system’s known involvement in circadian regulation and sleep-wake homeostasis provides a plausible mechanistic basis for expecting cannabinoids to influence sleep architecture. Gates and colleagues, publishing in Sleep Medicine Reviews (2025), conducted the first systematic review and meta-analysis restricted to polysomnographic studies, the gold standard for objective sleep measurement. By limiting their scope to PSG-derived outcomes, the authors aimed to resolve ambiguity left by prior reviews that relied on subjective sleep diaries or questionnaires and often conflated different cannabinoid preparations, doses, and populations.

Across 18 included studies (9 contributing to meta-analysis), pooled estimates showed no consistent effect of cannabis administration on total sleep time, sleep onset latency, wake after sleep onset, sleep efficiency, REM sleep percentage, or slow wave sleep. Earlier reports of THC-induced REM suppression, long cited in clinical teaching, were traced primarily to high-dose, small-sample trials from the 1970s with significant methodological limitations. In contrast, cannabis withdrawal emerged as the most robust and consistent finding: regular users who discontinued cannabis reliably experienced reduced total sleep time, prolonged sleep onset latency, and REM rebound. The authors emphasize that the evidence base is marked by profound heterogeneity in dose, cannabinoid composition, delivery method, population characteristics, and prior use history, and they caution that the null findings for administration should not be interpreted as evidence of safety or efficacy. They conclude that well-powered, standardized randomized controlled trials with uniform cannabinoid preparations are needed before any clinical recommendations can be made.

Dr. Caplan’s Take

This review is valuable precisely because it refuses to flatten a complex evidence base into a simple answer. Patients ask me regularly whether cannabis will help them sleep, and what I have to tell them is that the best objective data we have does not support a consistent benefit, even though many people feel subjectively that it works. That disconnect between perceived and measured sleep quality is something clinicians encounter often with sedating substances, and it deserves more attention. The mechanistic rationale is real, but the clinical proof is simply not there yet.

In practice, when a patient is using cannabis for sleep, I focus first on understanding their use pattern and duration, because the withdrawal data in this review is clinically meaningful. If someone has been using nightly for months and wants to stop, I counsel them explicitly about the likelihood of transient sleep disruption, including vivid dreaming from REM rebound, and I help them plan a gradual taper with concurrent behavioral sleep strategies. I do not recommend initiating cannabis for sleep based on current evidence, and I am transparent with patients about why.

Clinical Perspective

This review sits early in the research arc for cannabis and sleep, despite the long history of cannabinoid pharmacology. Its primary contribution is demonstrating that the objective evidence does not support the popular narrative that cannabis reliably improves sleep architecture. Clinicians should note that the earlier literature suggesting robust REM suppression from THC has not been replicated at therapeutic doses in more methodologically sound studies. The withdrawal findings, however, are clinically solid and should inform cessation counseling for any patient using cannabis regularly. The evidence does not currently support recommending cannabis, in any formulation, as a first-line or adjunctive sleep intervention.

From a pharmacological standpoint, the review’s inclusion of THC, CBD, nabilone, dronabinol, and combination products like nabiximols (Sativex) underscores that “cannabis” is not a single intervention, and treating it as one has been a persistent source of confusion. CBD and THC may have opposing effects on arousal and sleep staging, and dose-response relationships remain poorly characterized. Drug interaction potential with sedative-hypnotics, benzodiazepines, and anticonvulsants should be considered in polypharmacy contexts. The single most actionable recommendation clinicians can implement now is to proactively counsel regular cannabis users about the well-documented sleep disruption associated with abrupt cessation and to incorporate structured tapering plans when discontinuation is clinically indicated.

Study at a Glance

Study Type

Systematic review and meta-analysis

Population

Cannabis-naive individuals, chronic cannabis users, and patients with comorbidities including PTSD, chronic pain, Parkinson’s disease, insomnia, and obstructive sleep apnea

Intervention

Cannabis or cannabinoid administration (THC, CBD, nabilone, dronabinol, nabiximols) via smoked, vaporized, oral, or oromucosal delivery

Comparator

Placebo or baseline (pre-withdrawal) conditions, depending on study design

Primary Outcomes

PSG-measured total sleep time, sleep onset latency, wake after sleep onset, sleep efficiency, REM sleep percentage, and slow wave sleep

Sample Size

18 studies included (9 contributing quantitative data to meta-analysis)

Journal

Sleep Medicine Reviews, Volume 84 (2025)

Year

2025

DOI or PMID

Not available in extracted text

Funding Source

Not specified in extracted text

What Kind of Evidence Is This

This is a pre-registered systematic review and meta-analysis, which occupies a high position in the evidence hierarchy when the underlying primary studies are methodologically strong and reasonably homogeneous. In this case, the most important inference constraint is that only 9 of 18 identified studies contributed data suitable for quantitative pooling, and those studies varied substantially in dose, cannabinoid type, delivery route, population, and design. This limits the statistical power and clinical generalizability of the pooled effect estimates considerably.

How This Fits With the Broader Literature

This review aligns with and extends prior narrative reviews that noted inconsistency in the cannabis-sleep literature but lacked quantitative synthesis of PSG data specifically. A 2017 review by Babson, Sottile, and Morabito similarly concluded that evidence for cannabinoids as sleep aids was preliminary, though that analysis relied more heavily on subjective outcomes. The current review’s restriction to polysomnographic data represents a methodological advance and helps explain the persistent discrepancy between patients’ subjective experience and objective measurement. The withdrawal findings are consistent with a well-established body of literature on cannabis withdrawal syndrome, which was formally recognized in DSM-5 and includes sleep disturbance as a core feature.

Common Misreadings

The most likely overinterpretation is concluding that this review proves cannabis has no effect on sleep. Absence of a consistent pooled effect across a small, heterogeneous evidence base is not the same as evidence of no effect. It is entirely possible that specific cannabinoid formulations at specific doses in specific populations do meaningfully alter sleep architecture, but the current literature is too fragmented and underpowered to detect or confirm such effects. Equally problematic would be citing this review to argue that cannabis is therefore safe for sleep, since the lack of demonstrated benefit does not equate to an established safety profile for chronic nightly use.

Bottom Line

This systematic review and meta-analysis establishes that cannabis administration does not consistently alter objective sleep architecture based on available polysomnographic evidence. It does not establish that cannabis is ineffective, safe, or harmful for sleep. The most clinically actionable finding is that cannabis withdrawal reliably disrupts sleep, which should directly inform cessation counseling. Larger, well-controlled trials with standardized cannabinoid preparations are needed before any evidence-based sleep recommendations can be made.

References

1. Gates PJ, Albertella L, Copeland J. The effects of cannabinoid administration on sleep: a systematic review of human studies. Sleep Medicine Reviews. 2025;84. Published in Sleep Medicine Reviews, Volume 84.
2. Babson KA, Sottile J, Morabito D. Cannabis, cannabinoids, and sleep: a review of the literature. Current Psychiatry Reports. 2017;19(4):21.
3. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 5th ed. Arlington, VA: American Psychiatric Publishing; 2013.