REM Sleep vs. Deep Sleep: What's the Real Difference?

What this actually measures

Neither stage is something you can feel from the inside. Both are scored from polysomnography — a monitored sleep study that records brain-wave activity (EEG), eye movement and chin muscle tone alongside heart rate and breathing. MedlinePlus describes that same study measuring how long it takes to fall asleep and how long it takes to reach REM specifically, which is how a lab tells REM's fast, wake-like waveform and paralysed muscles apart from deep sleep's slow, high-amplitude waves and comparatively normal muscle tone.

Two different brain states, not two flavours of “asleep”

Deep sleep and REM sleep aren’t points on the same dial, turned up or down. They’re close to opposites on the measures that actually define a sleep stage: what the brain’s electrical activity looks like, what the muscles are doing, and what the eyes are doing.

Deep sleep — the third and final non-REM stage, usually called slow-wave sleep — is defined by a particular slow, high-amplitude pattern in brain activity, according to the NIH’s National Heart, Lung, and Blood Institute (NHLBI). The body is at its most physically still here. It’s the stage most people mean when they describe a night as having been genuinely restful, even if they couldn’t tell you why.

REM sleep is close to the opposite signature. Brain activity picks up and starts to resemble wakefulness. The eyes twitch beneath closed lids — hence the name, rapid eye movement — and the major muscles go slack under a temporary paralysis that stops the body from acting out whatever the brain is doing. Most dreaming happens here. It is, by every measure that defines it, a more active state than deep sleep, not a lighter version of it.

That’s the first correction worth making, because “light sleep, REM sleep, deep sleep” gets talked about as a ladder, with deep sleep at the bottom and REM somewhere in the middle. It isn’t a ladder. It’s two different jobs, done by two different brain states, on two different schedules.

Where each one actually sits across the night

Sleep runs in cycles, and NHLBI puts a number on the length: roughly 80 to 100 minutes per cycle, with four to six cycles across a typical night. Each cycle moves through the non-REM stages before reaching REM, then starts again.

The two stages aren’t spread evenly through those cycles, though. NHLBI describes deep sleep as concentrated earlier in the night — you get more of it in the first cycle or two — while REM becomes more prominent later, closer to the morning. Practically, that timing has a consequence that most sleep-hygiene advice skips over.

If a night gets shortened at the back end — a normal bedtime followed by an early alarm, which is the most common way sleep actually gets cut short on a workday — the stage that disproportionately gets cut is REM, because that’s the stage concentrated in the hours you just removed. Deep sleep, sitting earlier in the night, is comparatively protected by the same schedule that leaves REM exposed. A late, rushed bedtime does the opposite: it can compress the early cycles and cut into deep sleep before REM has had its turn.

That’s a reason to think about which end of a short night got shortened, rather than just totting up hours and assuming the deficit is spread evenly across every stage.

How many stages of sleep are there, actually

Four, not two and not five. MedlinePlus describes non-REM sleep as having three stages that “range from light to deep sleep,” with REM sleep as the fourth, separately defined stage. NHLBI names that third non-REM stage specifically as deep, or slow-wave, sleep — the destination at the “deep” end of that light-to-deep range.

So the shorthand “REM vs. deep sleep,” useful as it is for comparing the two most talked-about stages, skips over stages one and two of non-REM sleep entirely. Those earlier stages aren’t padding. NHLBI describes stage 1 as the transition between wakefulness and sleep, and stage 2 as the point at which you are, technically, asleep — before the cycle has even reached the deep-sleep stage people mean when they use the phrase. A full night moves through all four, repeatedly, not through a subset of favourites.

Is there a “right” percentage of REM sleep?

Here’s the honest answer, and it’s less satisfying than a number: there isn’t one published target percentage to aim for. NIH material on sleep stages describes how the proportion of REM shifts — more of it later in the night, more of it in infancy than in adulthood — rather than naming a figure a healthy adult should be hitting on any given night.

That’s worth sitting with, because plenty of consumer sleep-tracking content states a specific REM percentage as though it were an official target. If a number like that isn’t attached to a named institutional source measuring it that way, treat it as an estimate from whatever device or app is doing the reporting, not as a clinical benchmark. What NHLBI does document with confidence is the shape of the pattern across a night and across a life — and that pattern, not a target percentage, is the part actually worth acting on: protecting the later hours of a night protects REM specifically, in the same way protecting the early hours protects deep sleep.

Which one is more important?

Neither, on the evidence that’s actually published. MedlinePlus states this plainly for sleep stages generally: you need all of them to get healthy sleep. Not more of the favourite one — all of them, including the two lighter non-REM stages that rarely get discussed at all.

That’s not evasiveness. It reflects that deep sleep and REM sleep were shown, by the same descriptions above, to be doing genuinely different things — one physically still with slow-wave brain activity, one behaviourally paralysed but neurologically active — rather than one being a stronger or weaker version of the other.

It’s also worth stating the part that complicates the popular idea that REM is the “restorative for the mind, gentle on the body” stage. NHLBI notes that the sympathetic nervous system — the body’s activating branch — switches on during REM sleep and on waking, raising heart rate and blood pressure. REM is not cardiovascularly quiet. That doesn’t make it a problem stage; it makes “deep sleep is for the body, REM is calm and just for the mind” a tidier story than the physiology actually supports.

How this gets measured, and why a wearable’s graph isn’t the same thing

Polysomnography — a monitored, overnight sleep study — is the actual measurement tool, and MedlinePlus is specific about what it records: brain activity through scalp electrodes, eye movement, muscle tone at the chin, plus heart rate and breathing. The same study measures how long it takes to fall asleep and how long it takes to reach REM specifically, which is exactly the pair of signals that separate REM’s fast, wake-like waveform from deep sleep’s slow one.

That same test is what’s used to diagnose sleep apnea, narcolepsy, periodic limb movement disorder, and REM sleep behavior disorder — conditions that show up as a disruption to this specific stage architecture, not just as “bad sleep” in general.

A phone or wrist-worn sleep tracker is not running that test. It is inferring stage from movement and heart-rate patterns, which correlate with sleep stages but are not the same signal a scalp electrode reads directly. That’s a reasonable estimate for spotting a trend in your own data over weeks. It is not the tool that diagnoses REM sleep behavior disorder or narcolepsy, and a single night’s stage breakdown from a consumer device shouldn’t be read with the same confidence as a clinical study.

What changes with age

Two separate trends get conflated under “sleep gets lighter as you age,” and NHLBI actually describes them as distinct. REM sleep is proportionally highest in newborns. Slow-wave, deep sleep follows its own arc: it peaks in early childhood, drops during the teenage years, and continues declining through adulthood — to the point that NHLBI notes older adults may get very little of it, or none.

That means an older adult’s sleep isn’t uniformly “worse” across every stage. It’s specifically lighter on the deep-sleep end, following a decline that starts well before old age and runs continuously from the teenage years onward, while REM follows a different lifelong curve entirely. Two trends, two different starting points, one shared label.

Where a stage problem stops being something to read about

Most of what varies night to night in your own REM and deep sleep is not a medical question — it shifts with bedtime, wake time, and how the night before went, and none of that needs a clinician. Two patterns are different.

Physically acting out a dream — punching, kicking, shouting, or ending up on the floor while still asleep — is a marker of REM sleep behavior disorder specifically, and it’s on the list of conditions polysomnography is used to diagnose, per MedlinePlus. That pattern is worth a monitored sleep study, not a change in your evening routine.

Recurring sleep paralysis alongside sudden muscle weakness or vivid hallucinations right at sleep onset is a different real pattern — REM activity intruding into wakefulness — and it’s also worth describing to a clinician directly rather than researching further on your own. Both are cases where the stage architecture itself, not just the total hours, has gone wrong, and that’s a different problem than the one this guide can answer.