Short answer: cannabis makes most people fall asleep faster, and that part is real — but the evidence that it improves sleep is much weaker than the evidence that it sedates you. Reviews of the research find that THC reduces the time it takes to fall asleep while appearing to impair sleep quality with long-term use (Babson et al., 2017, Current Psychiatry Reports). The gap between “I fell asleep quickly” and “I slept well” is where most of the complication lives.
This is not a moral question. It is a mechanism question, and the mechanism is genuinely interesting.
What does THC actually do to sleep?
Falling asleep faster is the most consistent finding. THC has sedative properties at low to moderate doses, and for someone lying awake with a busy mind, that effect is immediate and obvious. It is the reason cannabis feels like it works far more convincingly than most supplements do.
What happens after you fall asleep is less flattering. The literature consistently reports REM suppression — less time in the dreaming stage of sleep — along with, in some studies, an increase in slow-wave sleep early in the night that fades with repeated use (Babson et al., 2017, Current Psychiatry Reports). If you have ever noticed that you stop remembering dreams entirely when you use nightly, that is the REM suppression showing up in your own experience.
Suppressed REM is not automatically harmful for a night or two. But REM is where a large share of emotional processing and memory consolidation happens, and it is one of the stages that gets shortchanged first when anything chemical is on board. We walk through what each stage does in sleep stages explained.
The population-level picture is the most quietly revealing. In an analysis of over 21,000 adults from NHANES, recent cannabis users had higher odds of reporting both short sleep (under six hours) and long sleep (over nine), with heavy users — 20 or more of the past 30 days — showing the strongest association (Diep et al., 2022, Regional Anesthesia & Pain Medicine). Not better sleep. More extreme sleep, in both directions.
Why does it stop working?
Tolerance to the sedating effect develops faster than tolerance to almost everything else cannabis does. This is the pattern nearly everyone who uses nightly eventually describes: it worked beautifully for a few weeks, then it stopped working, then the dose went up, then it stopped working again.
The same thing happens with alcohol as a sleep aid, and for structurally similar reasons. A substance that sedates you into sleep is not the same as a substance that improves the sleep you get, and the body adapts to sedation quickly.
What makes cannabis stickier than alcohol for sleep specifically is what happens when you stop.
What happens when you stop using?
Sleep difficulty is one of the most reliable and longest-lasting features of cannabis withdrawal. In the review that established cannabis withdrawal as a valid clinical syndrome, sleep problems and strange, vivid dreams appeared among the core symptoms, typically peaking in the first week and sometimes persisting for several weeks (Budney et al., 2004, American Journal of Psychiatry).
This has been measured directly in a sleep lab. Heavy cannabis users studied during abstinence showed worse sleep efficiency, less total sleep time, more time awake after falling asleep, and shorter REM latency than drug-free controls (Bolla et al., 2008, Sleep).
Here is why that matters more than it might sound. If stopping temporarily makes sleep worse, then a few bad nights become powerful evidence that you need it — when what you are actually experiencing is the rebound. Researchers have flagged this loop explicitly, because disrupted sleep during abstinence is one of the strongest predictors of returning to use.
That is not a reason to feel bad about it. It is a reason to expect the first one to two weeks to be rough, rather than reading them as proof of anything.
Does this work differently for new parents?
For breastfeeding parents, the answer is unusually clear for a sleep topic. When researchers analyzed breast milk from mothers who used cannabis, THC was detectable in roughly 63% of samples, at a median concentration of 9.47 ng/mL, for up to about six days after last reported use (Bertrand et al., 2018, Pediatrics). THC is fat-soluble and lingers; there is no “pumped and dumped it, we’re fine” window that the data supports. Both the American Academy of Pediatrics and ACOG advise against use while nursing.
Separately, there is a practical issue that has nothing to do with milk. Nightly sedation makes it harder to surface from sleep, and new-parent nights are built around surfacing from sleep. Fragmented nights already blunt the recovery you get from the hours you do sleep, and adding a sedative on top of that tends to make the mornings heavier rather than lighter.
If you are postpartum and lying awake even when the baby is asleep — which is its own distinct and common problem — the postpartum insomnia research points at different interventions entirely. And if the underlying issue is simply not enough total sleep, that is a scheduling problem more than a sleep-quality problem, which we cover in sleep debt for parents.
Why are so many perimenopausal women using cannabis for sleep?
Because the symptom is real and the standard options have often felt unavailable.
In a survey of 258 perimenopausal and postmenopausal participants, 86.1% reported current cannabis use, and 78.7% endorsed using it specifically for menopause-related symptoms. The single most commonly cited symptom was sleep disturbance, at 67.4%, followed by mood and anxiety at 46.1% (Dahlgren et al., 2022, Menopause). That is a striking number, and it says less about cannabis than it does about how many women reached midlife, found their sleep coming apart, and were not offered much.
The catch is mechanistic. Perimenopausal sleep disruption is largely driven by falling and erratic progesterone and estrogen, and by night-time temperature surges that fragment sleep from the outside in. A sedative does not act on either of those. It can make falling asleep easier while the hot flash that wakes you at 3am still wakes you at 3am — you are just groggier when it does.
The interventions with actual evidence for this pattern are hormonal and temperature-focused. We go through them in perimenopause and sleep and night sweats and sleep.
Is CBD any different?
Somewhat, mostly in that it does less. CBD is non-intoxicating and does not produce the same sedation or REM suppression, but the randomized-trial evidence for it as a sleep aid is thinner than the packaging suggests — when trials are pooled, the formulations that improve subjective sleep tend to be the ones containing THC or CBN, not CBD alone. We covered that literature in detail in does CBD help you sleep.
What has better evidence?
For chronic difficulty sleeping, cognitive behavioral therapy for insomnia outperforms every pharmacological option in long-term outcomes and is the first-line recommendation in major clinical guidelines. It is unglamorous, it takes a few weeks, and it keeps working after you stop doing it — which is precisely the property cannabis lacks. The components are laid out in how to treat insomnia without medication.
Before that, the cheap structural things: a consistent wake time, morning light, and a cool room. They are boring, and they compound.
The calm version
Cannabis genuinely does shorten the time it takes to fall asleep — that experience is not imagined, and noticing it does not mean you have been fooling yourself. What the research suggests is that the effect fades with regular use, that REM sleep takes the hit, and that the difficult nights after stopping are rebound rather than proof of need. If you are using it nightly and it has stopped working, that is the expected pattern, not a personal failing. Mendtide’s coach works from your own sleep data rather than a rulebook, and it cites the research behind what it suggests, so you can see the reasoning instead of being told what to do.
Falling asleep faster and sleeping better are two different things. Most sleep aids are good at the first one. It’s worth knowing which one you actually needed.