Sleep Stages Explained: N1, N2, N3, and REM

Sleep is not one thing. It’s a sequence of distinct brain states, each with its own biological signature and function. Understanding the four stages, what they do, when they happen, and why they matter, is the foundation for everything else in sleep science.

This guide covers the modern four-stage model used by sleep researchers and clinicians, what happens in each stage, and how they assemble into the sleep cycle that shapes your night.

The four sleep stages

The current classification, established by the American Academy of Sleep Medicine in 2007, identifies four stages:

Stage	Type	% of night	Per cycle	Key feature
N1	Non-REM, light	<5%	1–7 min	Transition from waking
N2	Non-REM, light	45–55%	10–25 min	Stable sleep, spindles
N3	Non-REM, deep	13–23%	20–40 min (early)	Slow-wave, restorative
REM	REM	20–25%	10–60 min	Dreams, paralysis

You’ll sometimes see older literature reference “stages 1–4” plus REM. That five-stage system was used until 2007. Stages 3 and 4 (both slow-wave sleep) were merged into N3 because they’re functionally similar, both are deep sleep.

N1: the transition zone

N1 is the shallowest stage of sleep, the brief interval as you drift from waking into sleep proper. It typically lasts just a few minutes at the start of the night and even shorter durations between later cycles.

What’s happening:

• Brain waves slow from beta (alert waking, 12–30 Hz) to alpha (relaxed, 8–12 Hz) and then to theta (4–8 Hz).
• Muscles relax. Some people experience hypnic jerks, involuntary muscle twitches that can wake them. These are normal and harmless.
• Heart rate and breathing slow slightly.
• You may experience brief, dream-like imagery called hypnagogic hallucinations.

You’re easily woken from N1, and you might not even realize you were asleep if someone wakes you. If you’ve ever “drifted off” during a meeting or a boring lecture and then snapped back to awareness, you were probably in N1.

N1 makes up less than 5% of total sleep across a typical night. Its main job is to transition you in and out of deeper sleep.

N2: the workhorse stage

N2 is where you spend the largest share of your sleep, about half the night across all cycles. It’s still “light” sleep compared to N3, but you’re considerably more disconnected from the environment than in N1.

What’s happening:

• Sleep spindles appear: brief bursts of high-frequency brain activity (12–14 Hz) lasting about a second. These are thought to play a role in memory consolidation and protecting sleep from external disturbances. People with more spindles tend to be better learners.
• K-complexes appear: large, single brain-wave events that can occur in response to external stimuli or spontaneously. They may suppress arousal, keeping you asleep despite noise or other disturbances.
• Body temperature drops further. Heart rate and breathing slow more.
• Muscles relax further, though you can still move.

N2 sleep is when most of your declarative memory consolidation happens, facts, names, vocabulary, procedures. It’s also a deep enough stage that brief environmental disturbances usually don’t wake you, but light enough that genuine emergencies can.

A typical 20-minute “power nap” mostly stays in N2, which is why power naps work. You get the alertness benefit of brief sleep without entering N3 and risking sleep inertia.

N3: deep slow-wave sleep

N3 is the deepest and most physically restorative stage of sleep. It’s also the hardest to wake from, and waking from it causes the worst sleep inertia.

What’s happening:

• Delta waves dominate the EEG: large, slow waves at 0.5–4 Hz with high amplitude. These reflect synchronized firing of huge populations of cortical neurons.
• Growth hormone surges. The largest pulse of nightly growth hormone release occurs during the first N3 phase.
• Glymphatic clearance accelerates. The brain’s waste-removal system runs at peak efficiency, flushing beta-amyloid and other metabolic byproducts.
• The immune system rebuilds. Cytokines and other immune signaling proteins are produced predominantly during N3.
• Blood pressure and heart rate are at their lowest.

N3 is heavily front-loaded in the night. You get about 70% of your total N3 in the first two sleep cycles (the first 3 hours of sleep). Later cycles have little or no N3.

This is why short sleep that starts at your usual bedtime preserves most of your deep sleep, while pushing your bedtime later eats directly into N3. It’s also why staying up late and “sleeping in” doesn’t fully recover the deep sleep you missed, late-morning sleep is mostly N2 and REM.

For a deeper look at N3 specifically, see our deep sleep guide.

REM: rapid eye movement sleep

REM is the strange one. The brain is highly active, almost as active as alert waking. The eyes move rapidly under closed lids. Most major muscles are paralyzed to prevent acting out dreams. And the most vivid, narrative dreams happen here.

What’s happening:

• EEG looks like alert waking. High-frequency, low-amplitude activity dominates.
• Atonia. Most skeletal muscles are paralyzed via a brain-stem mechanism. The eyes and diaphragm are exceptions.
• Autonomic instability. Heart rate and blood pressure become variable. Body temperature regulation pauses.
• Sexual arousal. Penile or clitoral engorgement is part of normal REM in healthy adults, regardless of dream content.
• Eyes move rapidly. This is the namesake feature, discovered in 1953 by Aserinsky and Kleitman.

REM is concentrated in the second half of the night. The first REM phase is brief (~10 minutes), the last can stretch to 60+ minutes. About 20–25% of total sleep is REM.

Functionally, REM appears to handle emotional memory consolidation, creative problem-solving, and emotional regulation. Selective REM deprivation in lab studies impairs all three. For more, see our REM sleep guide.

How the stages fit together

A typical sleep cycle goes:

N1 → N2 → N3 → N2 → REM → (brief micro-awakening) → repeat

Each cycle lasts roughly 90 minutes, with significant individual variation (70–120 minutes is typical). A healthy adult completes 4–6 full cycles per night.

The cycles aren’t identical. Early cycles look like:

• Cycle 1: N1 (5 min) → N2 (15 min) → N3 (35 min) → N2 (15 min) → REM (10 min) = ~80 min
• Cycle 2: N1 (3 min) → N2 (20 min) → N3 (25 min) → N2 (15 min) → REM (15 min) = ~78 min

Late cycles look very different:

• Cycle 4: N1 (3 min) → N2 (30 min) → (no N3) → N2 (15 min) → REM (40 min) = ~88 min
• Cycle 5: N1 (3 min) → N2 (25 min) → REM (50 min) = ~78 min

This asymmetry has practical consequences. Late bedtimes cost you mostly N3 (front-loaded). Early wake-ups cost you mostly REM (back-loaded). “Sleeping in” gives you mostly REM. Short sleep that’s consolidated and starts on time gives you most of your N3.

To see this visually, try the sleep cycle calculator, which renders a hypnogram showing the predicted stages across your night.

How to think about stage time

You don’t directly control how much time you spend in each stage, your brain decides based on sleep pressure, circadian rhythm, and individual biology. But you do influence the conditions that affect stage distribution:

• Total sleep duration determines how many cycles you get. More cycles means proportionally more REM (since late cycles are REM-rich).
• Consistent timing keeps your circadian rhythm aligned, which improves the efficiency of each stage.
• Sleep environment affects fragmentation. A cool, dark, quiet room favors deep N3.
• Alcohol, caffeine, and medications can suppress specific stages, usually REM and/or N3.
• Age strongly affects stages. Children have massive amounts of N3; older adults have very little.

What about sleep tracker data?

If you wear a sleep tracker (Oura, Whoop, Apple Watch, Fitbit, Garmin, etc.), it will report nightly time in each stage. These estimates are useful for tracking trends, your average deep sleep over a month, whether late drinking reliably tanks your REM, but treat them as rough.

Consumer wearables don’t measure EEG. They use heart rate, heart rate variability, breathing rate, and movement to infer stages. Validation studies show:

• Total sleep time is usually accurate to within 5–10 minutes.
• REM detection is moderately accurate (60–80% agreement with polysomnography).
• Deep sleep detection is the least accurate, most trackers overestimate, sometimes significantly.

Don’t obsess over the absolute numbers. Use the trends. If your deep sleep drops from your usual baseline when you drink wine with dinner, that pattern is reliable even if the absolute “32 minutes deep” figure isn’t precise.

The takeaway

The four sleep stages, N1, N2, N3, REM, aren’t equal. They serve different functions, are distributed unevenly across the night, and are affected differently by lifestyle. Knowing this lets you make smarter decisions about when to sleep, how long, and what to avoid.

If you remember only one thing: stage distribution is the why behind the 90-minute sleep cycle’s importance. The reason waking at the end of a cycle feels gentle and waking mid-cycle feels brutal is that you’re waking from light sleep vs. N3 deep sleep, two completely different brain states. Aligning your alarm to cycle endings is the easiest sleep optimization there is.