NREM vs REM Sleep: What Actually Happens in Each Stage

NREM vs REM Sleep: What Actually Happens in Each Stage — Proco
NREM vs REM Sleep: What Actually Happens in Each Stage — Proco

Sleep isn't one state — it's four distinct stages your brain cycles through roughly every 90 to 110 minutes, four to five times a night: three stages of non-REM sleep (N1, N2, N3) and REM.3 Each stage does different physiological work. N3, often called deep or slow-wave sleep, is where the body does most of its physical recovery, and it's concentrated in the first half of the night. REM is where the brain does more of its memory consolidation and emotional processing, and it dominates the second half, getting longer with each cycle toward morning.12 This matters for anyone dealing with insomnia, because poor sleep isn't just "not enough hours" — research on sleep architecture shows insomnia can specifically disrupt the balance and continuity of these stages, not just total sleep time.4 It also matters for anyone taking a sleep supplement: most sleep ingredients, melatonin included, are studied for getting you to sleep faster, not for reshaping how much N3 or REM you actually get once you're there.5 Here's what happens in each stage, and where the supplement evidence does — and doesn't — reach.

Evidence reviewed: September 2026.

Corrected September 2026: two claims overstated their sources. The insomnia REM study was described as finding that REM fragmentation "tracked with the gap between how poorly people felt they'd slept and what the overnight recording actually showed." Its subtypes were defined by polysomnography-detected structural abnormality, not by a subjective-objective mismatch measure, and the paper reports no such correlation analysis — that framing has been removed and replaced with what it actually found, along with its sample size and journal. And the melatonin network meta-analysis was credited with finding a "moderate-to-large reduction in sleep onset latency (physiologically measured)"; the review reports support for melatonin on sleep-onset difficulties without those qualifiers, so they have been dropped and its headline conclusion about CBT-I added instead.

The four stages, one at a time

Sleep researchers score a night's sleep using polysomnography — simultaneous EEG, eye-movement, and muscle-tone recording — and classify it into four stages under the current American Academy of Sleep Medicine (AASM) system.3

N1 — light sleep, the on-ramp. Lasts one to a few minutes per cycle. Brain waves slow slightly, muscle tone starts to ease, and it's the easiest stage to be woken from. It typically makes up around 5% of total sleep time.2

N2 — the largest share of the night. Body temperature drops, heart rate and breathing slow, and eye movement stops. EEG recordings show sleep spindles and K-complexes — brief bursts of brain activity thought to help consolidate memory and keep you asleep despite background noise. N2 makes up roughly 45–50% of a typical night's sleep, and each N2 period gets longer as the night goes on.12

N3 — deep or slow-wave sleep. Marked by large, synchronized delta brain waves. Muscle activity, pulse and breathing all drop further, and this is the hardest stage to wake someone from — people woken here often feel groggy for a period afterward (sleep inertia). N3 is concentrated in the first third to half of the night and shrinks with each successive cycle, which is part of why a short night costs you deep sleep disproportionately. We've gone deeper on what does and doesn't change it in the science of deep sleep.12

REM — the stage most associated with dreaming. Brain activity on EEG looks close to waking activity, even though the body is in temporary muscle paralysis (atonia) apart from the eyes and breathing muscles — hence "rapid eye movement." REM periods lengthen across the night and are longest in the final cycles before you wake, making up roughly 20–25% of total sleep time in a typical adult night.12 That atonia is not incidental: when it fails, people physically act out dreams, which is a recognised clinical picture rather than a quirk — covered in sleepwalking, night terrors and acting out dreams.

4–5
Full sleep cycles a typical adult moves through in one night, each roughly 90–110 minutes, following the pattern N1 → N2 → N3 → N2 → REM — with deep sleep concentrated early and REM concentrated late.1

Why the order matters more than people think

Because deep sleep is front-loaded and REM is back-loaded, when you cut your sleep short changes what you actually lose. A late bedtime that still lets you sleep a full morning tends to cost you more N3; an early alarm that cuts your last hour or two tends to cost you more REM, since that's when REM periods are longest.1 Neither is "worse" across the board — they're doing different jobs — but it's a useful reason to think about sleep architecture as a shape, not just a number of hours, and a useful frame for how much sleep adults actually need.

What insomnia actually does to that shape

Insomnia doesn't just mean less total sleep. A study in Sleep Medicine analysing 1,049 participants — 141 good sleepers, 214 people with chronic insomnia but no conventional polysomnographic abnormality, and 694 with chronic insomnia and PSG-detected structural abnormality — found that both insomnia groups showed more REM arousals and awakenings per hour, higher REM density, and a greater proportion of phasic REM than good sleepers.4 The notable part: those REM abnormalities were present even in the group whose standard sleep-architecture measures looked normal. A model using REM features distinguished the two insomnia subtypes with an area under the curve of 0.752 — useful, not definitive. The authors' conclusion was that REM microstructure captures clinically relevant aspects of insomnia that standard sleep architecture measures don't reflect.

Separately from that study, the mismatch between how badly people feel they slept and what a recording shows is a real and well-documented feature of chronic insomnia — people can wake convinced they barely slept when a recording shows otherwise. Whether REM instability explains that mismatch is a reasonable hypothesis rather than something this research established, and we'd rather label it as such. Either way, it's a reminder that "sleep isn't restful" and "sleep is short" are related but genuinely different problems — and a bottle of anything is unlikely to fix both the same way. If unrefreshing sleep is your pattern, the first thing worth ruling out isn't a supplement at all but a breathing problem: the STOP-BANG screening questions take two minutes.

Where sleep supplements fit — and where they don't

Most ingredients marketed for sleep, including melatonin, have their strongest trial evidence for sleep onset — how long it takes you to fall asleep — not for changing the proportion of N3 or REM you get once you're under. A 2020 systematic review and network meta-analysis in the Journal of Clinical Medicine, covering 40 studies of melatonin, light exposure, exercise and complementary interventions for insomnia disorder (36 entering the meta-analysis), found the evidence supported melatonin for improving sleep-onset difficulties, while support for self-reported sleep efficiency and insomnia severity went to meditative movement therapies rather than melatonin.5 Its overall conclusion is the one worth carrying: none of the interventions it examined received superior evidence to cognitive behavioural therapy for insomnia. Nothing in that evidence base shows melatonin, valerian, or the other botanicals in our range measurably rebalancing NREM/REM proportions — that's simply not what these ingredients have been studied for.

That's an honest boundary worth stating plainly. Sleep Formula discloses every ingredient's dose — valerian 150mg, chamomile 100mg, GABA 100mg, L-tryptophan 100mg, lemon balm 100mg, passion flower 100mg and melatonin 2mg — and each has trial support for sleep onset or calming the nervous system before bed; the full breakdown is here. Sleep Support pairs magnesium, B6 and 10mg of melatonin with a 14-botanical blend whose per-ingredient amounts are not disclosed, and which our manufacturer confirmed in September 2026 cannot be published. Neither product is built or marketed as something that changes your sleep architecture once you're asleep — the honest claim is "may help you get to sleep," not "will give you more deep sleep or more REM." And on melatonin specifically, the dose-response data puts the sleep-onset peak at around 4mg, which is above Sleep Formula's dose and below Sleep Support's — another reason not to read either as optimised.

Key takeaway: The stage of sleep you're in changes across the night in a predictable pattern — deep sleep early, REM late. Insomnia can disrupt that pattern's continuity, not just its length, and REM instability shows up even when standard architecture measures look normal. Sleep supplements' strongest evidence is for falling asleep faster, not for rebalancing the stages themselves; consistent sleep timing and total time in bed remain the biggest levers on the shape of your night.

Bottom line

Sleep isn't a single dial you turn up or down — it's four stages doing different jobs, cycling in a predictable order across the night. Deep sleep (N3) does more of the physical-recovery work and is concentrated early; REM does more of the memory and emotional-processing work and is concentrated late. Chronic insomnia research increasingly looks at REM instability specifically, not just total sleep time, as part of what makes sleep feel unrefreshing. And the supplement evidence that does exist — including for melatonin — is squarely about sleep onset, not about reshaping the architecture of the night. If insomnia or unrefreshing sleep is a persistent pattern for you rather than an occasional bad night, cognitive behavioral therapy for insomnia (CBT-I) has considerably stronger evidence than any supplement — we cover the evidence for it here, including the honest caveat that its own effect on total sleep time is small.

This isn't medical advice — talk to your doctor, especially if sleep problems are severe or persistent. Supplements are not intended to diagnose, treat, cure or prevent any disease.

References

  1. Stages of Sleep: What Happens in a Normal Sleep Cycle? Sleep Foundation. sleepfoundation.org
  2. Physiology, Sleep Stages. StatPearls, NCBI Bookshelf. ncbi.nlm.nih.gov
  3. The AASM Manual for the Scoring of Sleep and Associated Events. American Academy of Sleep Medicine. aasm.org
  4. REM sleep fragmentation and arousability distinguish chronic insomnia subtypes and reflect subjective-objective sleep discrepancy. Sleep Medicine. pubmed.ncbi.nlm.nih.gov
  5. Baglioni C, et al. A Systematic Review and Network Meta-Analysis of Randomized Controlled Trials Evaluating the Evidence Base of Melatonin, Light Exposure, Exercise, and Complementary and Alternative Medicine for Patients with Insomnia Disorder. Journal of Clinical Medicine, 2020;9(6):1949. pmc.ncbi.nlm.nih.gov

Related reading: The Science of Deep Sleep: Separating What Works from What's Marketing, Melatonin: The Real Dose-Response Evidence, Sleep Formula vs. Sleep Support: How to Actually Choose and CBT-I: The Sleep Treatment With Better Evidence Than Any Supplement.

Sleep Formula discloses every ingredient's dose — no blend, no guessing. Shop Sleep Formula → · Sleep Support pairs magnesium, B6 and 10mg melatonin with a 14-botanical blend. Shop Sleep Support →