REM vs Deep Sleep: Which One Actually Matters More?
Asier G. Morato
·
·
7
min read
Cover image generated with AI — © Chubby Studio S.L.
KEY TAKEAWAY
Neither REM nor deep sleep is more important overall. When a night runs short, the body protects deep sleep first and loses REM from the final hours, but REM is paid back on recovery nights and lower REM over years is associated with higher mortality in older adults. The practical lever is enough total sleep at a steady time, which protects both stages.
Neither, and the question hides a choice you never actually get to make. Deep sleep and REM do different jobs, they are produced by different pressures, and they sit in different halves of the night. When a night runs short, your body decides the order for you: it protects deep sleep first and lets REM go. Your body has already answered the question for you, at least for tonight.
That answer only holds for one night, though. Over the following nights REM comes back, and over years, people who get less of it do worse on average. So the honest version is this: deep sleep wins the short-term triage, REM is not the expendable one, and the thing you actually control is whether there is enough night for both.
What does each one actually do?
The short version, since what each sleep stage actually does covers the mechanism properly. Deep sleep (N3, slow wave sleep) is the stage where brain waves slow right down and you are hardest to wake. REM is the stage where your brain is nearly as active as when you are awake, your muscles are switched off, and most vivid dreaming happens.
The best known division of labour comes from memory research. A major review in Physiological Reviews by Rasch and Born describes slow wave sleep as the stage most consistently linked to consolidating declarative memory, the facts and events kind, through repeated replay of what you learned during the day. REM has been linked to procedural and emotional memory, but the same review is blunt about how shaky that is: in humans the evidence for REM's role in memory is described as surprisingly inconsistent, and even the near absence of REM during treatment with some antidepressants does not lead to obvious problems forming memories.
That second point is worth sitting with, because most articles on this topic skip it. REM's job in human memory is far less settled than the popular version suggests. The popular split, deep sleep for the body and REM for the mind, is a useful sketch rather than a finding, and neither stage has a single job.
If you had to pick one tonight, which would your body pick?
Deep sleep, and it is not close. Deep sleep is driven by sleep pressure, the drive that builds the longer you are awake, and it is spent early, in the first two or three cycles. REM is tied more to your body clock and stretches out toward morning, which is why the longest REM period of the night is usually the last one before your alarm.
So when you cut a night short, you do not lose each stage evenly. A late bedtime eats into some deep sleep, but the body compensates by dropping into it harder. An early alarm takes the final hours, and those hours are mostly REM. In practice, a short night is mostly a REM cut, which is also why a five-hour night can show a perfectly respectable deep sleep figure. The longer version of that accounting is in how much deep sleep you actually need.
That is the body's own ranking under scarcity. It is a triage rule for one night, not a statement that REM matters less.
See your whole night, not one stage
FitWoody scores each night against your own sleep goal, your usual patterns and standard sleep-quality parameters, instead of ranking one stage against another. Comes with a 7-day free trial.
Download the app
★★★★★
4.5 · 732 ratings on the App Store
Does that mean REM is the expendable one?
No. The clearest evidence is what happens after you lose it. When people get a chance to recover, REM comes back as well as deep sleep. A 2022 review of recovery from chronic sleep restriction in Sleep Advances found that giving people more time in bed after a stretch of short nights increased total sleep, REM and slow wave sleep, and concluded that it is not evident that any one sleep stage is more important for recovery than another. Recovery nights pay back both, which is not what a system does with a stage it can live without.
The same review is a useful reality check on how long that takes. After five nights restricted to four hours, three full nights of recovery were not enough to bring cognitive performance back to baseline. One long lie-in on Saturday does not clear a week of short nights, a point sleep debt covers in more depth.
Losing REM specifically also seems to change how you react the next day. In a small fMRI study published in Frontiers in Behavioral Neuroscience, twenty healthy men slept in a lab. Some were woken every time they entered REM, which cut it from about 21 percent of the night to about 4, while the comparison group was woken just as often during non-REM sleep. After that night, the REM-deprived group reacted more strongly to threatening images, and the comparison group did not change. It is one small study in men only, so treat it as a pointer rather than a rule, but it fits the everyday experience of feeling emotionally raw after a short night.
Is low REM linked to long-term health?
There is a striking association, with heavy caveats. A 2020 study in JAMA Neurology followed 2,675 older men (average age 76) for a median of 12 years, with sleep measured at the start by full overnight polysomnography, the clinical standard rather than a wrist device. Every 5 percentage points less REM was associated with a 13 percent higher rate of death from any cause, after adjusting for other factors. The researchers repeated the analysis in a second, younger group of 1,386 men and women (average age about 51) followed for around 21 years and found a similar pattern, with a 17 percent higher rate per 5 points. People with less than 15 percent REM did worse across every cause of death they looked at.
Now the caveats, which the authors state themselves. This is observational, so it cannot show that low REM causes anything; it may be a marker of illness that was already there, and the authors say further studies are needed to know whether the link is causal. It was measured with polysomnography, not a watch. And it describes groups of people over decades. It is an association measured in a sleep lab, across groups, not a verdict on your Tuesday night.
Which one should you watch on your Apple Watch?
Neither, on any single night. A watch infers stages from movement and heart rate rather than brain activity, and it is much better at telling asleep from awake than at telling REM from deep. We went through the validation numbers in how accurate wearable sleep tracking really is, and separating the stages is the weakest part of what any wrist device does.
What a watch can do reasonably well is show you a trend in your own estimates. If your REM minutes have drifted down across a month, the useful questions are boring ones. Has your alarm moved earlier? Are you drinking more in the evening? Alcohol is one of the most reliable REM suppressants there is. Have your weeknights quietly shortened? Those are the things that move REM, and none of them is solved by targeting REM directly.
Read minutes rather than percentages where you can, and do not set one stage against the other. An earlier bedtime and an earlier alarm can leave your deep sleep untouched and cost you REM, so a schedule that looks like an improvement can be a trade.
So what actually protects both?
Two things, and neither is specific to a stage. The first is enough total sleep, because deep sleep takes the start of the night and REM takes the end, and only a full night has both ends. The second is timing: a steady wake time keeps your body clock and your sleep pressure arriving together, which is why a consistent schedule does more than chasing extra hours.
Get those two right and both stages get the room they need without you managing either one. That is the version we built into FitWoody's sleep analysis: it scores your night against your own sleep goal, your usual patterns and standard sleep-quality parameters, rather than ranking one stage against another. It informs. It does not diagnose.
If your REM or your deep sleep has dropped sharply, stayed down for weeks, and you are feeling it during the day, that is a conversation for a doctor rather than a new routine. Otherwise, protect the whole night. The ranking takes care of itself.
Frequently asked questions
Is REM or deep sleep more important?
Neither is more important overall. Deep sleep is protected first when a night runs short, so it wins the short-term triage, but REM is also paid back on recovery nights and lower REM over many years is associated with worse health outcomes in older adults. They do different jobs and you need enough total sleep for both.
How much REM sleep do you need?
In healthy adults REM usually makes up around a fifth to a quarter of the night, most of it in the final hours. No major guideline sets a REM target; the guidance is about total sleep, seven or more hours a night for most adults. Protecting the end of the night is what protects REM.
Why do I get more REM than deep sleep?
Because REM builds toward morning while deep sleep is spent in the first few cycles, a long night naturally adds more REM than deep sleep. Deep sleep also declines with age, and wrist devices estimate stages from movement and heart rate, so the split on your watch is an estimate worth reading as a trend.