A sleep study divides the recording into wake, three non-rapid eye movement stages called N1, N2, and N3, and rapid eye movement sleep, or REM. These are standardized descriptions of patterns recorded during a sleep study. They are not separate health scores, and there is no single stage percentage that every person should reach.
N1, N2, N3, and REM usually recur in changing sequences across a night. More N3 tends to occur earlier in the sleep period, while REM episodes generally become more prominent later. The timing is not a rigid 90-minute clock, and normal transitions and brief awakenings can change the pattern from one cycle or night to the next 1.
How sleep stages are measured
In a clinical polysomnogram, a technologist scores short segments of the recording using signals from the brain, eyes, and chin muscles. Electroencephalography (EEG) records electrical activity at the scalp, electrooculography (EOG) records eye movements, and electromyography (EMG) records muscle tone. Together, those signals distinguish wake, N1, N2, N3, and REM according to American Academy of Sleep Medicine rules 2.
Sleep stages are often described as EEG-defined states because the brain-wave pattern is central to scoring. EEG alone is not the whole test, however. Eye movements and chin tone are especially important for separating REM from wake and N1. A full polysomnogram may also record breathing, oxygen, heart rhythm, leg movement, sound, and video, but those signals answer other clinical questions 3.
| State | Main laboratory clues | What the label means |
|---|---|---|
| Wake (W) | Awake EEG features, eye blinks or reading eye movements, and normal or higher chin tone | The scored segment is awake, even if the person lies still with eyes closed |
| N1 | Low-amplitude, mixed-frequency EEG activity, often with slow eye movements | A light, transitional sleep state that can occur at sleep onset or after an arousal |
| N2 | Sleep spindles or K-complexes appear in the EEG | A stable NREM state that commonly accounts for a large part of adult sleep |
| N3 | Slow, high-amplitude EEG waves meet the scoring rule | Slow-wave sleep, often called deep sleep |
| REM (R) | Low-amplitude, mixed-frequency EEG activity occurs with rapid eye movements and low chin muscle tone | A distinct state in which vivid dreaming is common and normal muscle atonia limits movement |
These categories turn a continuous, changing biological process into a readable clinical summary. A person does not necessarily pass through every stage in a fixed order each time. Arousals, brief wake periods, and returns to N1 or N2 are part of real hypnograms.
What happens in each sleep stage?
N1 is a transition into sleep
N1 often appears as wakefulness gives way to sleep. Awareness of the room fades, slow eye movements may appear, and muscle tone is still present. Someone awakened from N1 may feel that they were not asleep.
N1 is not useless sleep, but a large N1 share on one study can reflect frequent arousals, limited total sleep, the unfamiliar laboratory setting, or another source of fragmentation. Its meaning comes from the rest of the recording and the person's symptoms, not from the percentage alone 4.
N2 has spindles and K-complexes
Sleep spindles are brief bursts of rhythmic EEG activity. K-complexes are large, distinctive waves that may occur spontaneously or in response to a stimulus. Their presence allows a technologist to identify N2 under the scoring rules 3.
Research links spindles and other NREM activity with aspects of learning and memory, but the evidence is more detailed than “N2 stores memories.” Different tasks have been associated with different sleep features, and memory processing can involve coordinated activity across NREM and REM sleep 5. An app's N2 or “light sleep” total does not measure how much a person learned.
N3 is slow-wave sleep
N3 is identified by prominent slow waves in the EEG. Slow-wave activity rises with time awake, is strongest earlier in a usual sleep period, and can increase during recovery sleep after sleep loss. This makes N3 part of the homeostatic regulation of sleep, not a fixed nightly allotment 6.
Slow-wave sleep is associated with several physiological processes. The rate of growth hormone secretion is often greatest during this state, and NREM slow waves are studied alongside memory-related activity 65. These associations do not mean that a particular N3 percentage proves adequate tissue repair, immunity, hormone balance, or physical recovery. Nor does a low consumer “deep sleep” result diagnose a deficiency in any of those functions.
REM is more than dream sleep
During REM, the EEG looks relatively active, the eyes make rapid movements, and skeletal muscle tone is normally very low. Vivid, story-like dreams are commonly reported after REM awakenings, but dreaming can also occur during NREM sleep.
REM has been studied in memory and emotional processing. Reviews support a role for sleep in these processes and discuss evidence involving REM, but they do not establish REM as a stand-alone emotional reset or prove that one nightly REM percentage is optimal 57. Low REM on an app cannot explain a person's mood, memory, or daytime symptoms by itself.
How stages change across a night
Sleep usually begins in NREM, progresses into deeper NREM, and later reaches REM. The balance changes as the sleep period continues:
- N3 is usually concentrated toward the earlier part of the sleep period.
- REM episodes are often shorter early and longer later.
- N2 recurs throughout the night and often occupies more time as N3 becomes less prominent.
- Brief awakenings and stage shifts can appear between or within cycles.
“About 90 minutes” is a population shorthand, not an alarm-setting formula. In a laboratory study that recorded repeated nights in younger and older adults, individual cycles varied substantially and were longer when they contained more slow-wave or REM sleep. A spontaneous wake period also changed cycle duration 1. Counting backward in 90-minute blocks cannot tell when an individual will enter or leave a stage.
The pattern also depends on when and how long someone sleeps. Cutting a sleep period short may remove later REM-rich time. Sleeping after unusual hours, recent sleep loss, or a night of repeated arousals can produce a different hypnogram without showing that the brain has lost the ability to enter a stage.
There is no universal stage percentage
Published “normal” percentages are group summaries, not personal requirements. They change with age, sex, study methods, the night being measured, and the population selected.
Age findings also depend on what researchers compare. Lifespan research generally shows large developmental changes and an age-related reduction in slow-wave sleep, with more fragmented sleep in later life 8. Yet a meta-analysis of 169 studies and 5,273 healthy adults scored under newer AASM criteria found more N1 and more awakenings with age, while age trends in N2, N3, and REM percentages were not statistically significant 9. The results are not interchangeable: one synthesizes changes across human aging, while the other estimates parameters in carefully selected healthy adult control groups.
Infants and children also have age-specific sleep organization and scoring considerations. An adult stage target should not be applied to a child. Across any age, a clinician interprets stage distribution alongside total sleep time, timing, arousals, breathing, movement, medicines, symptoms, and the reason for the study.
Why a stage report may look different
A stage breakdown can shift for many reasons:
- Prior sleep loss: recovery sleep can carry greater slow-wave activity, so more N3 after a short night is not evidence that sleep loss was beneficial 6.
- Alcohol: a 2025 systematic review and meta-analysis of 27 studies found that presleep alcohol delayed REM onset and reduced REM duration. Effects on total sleep time, sleep efficiency, and wake after sleep onset remained uncertain 10.
- Medicines and other substances: antidepressants, sedatives, stimulants, and other agents can affect REM, NREM, sleep latency, or arousals. Starting, stopping, or changing a medicine may matter, so the interpreting clinician needs an accurate list. Do not stop a prescribed medicine to change a stage result 411.
- Sleep apnea and other sources of arousal: repeated breathing events, limb movements, pain, noise, or another disturbance can fragment sleep and alter stage proportions. The stage pattern does not identify which cause is present 4.
- Timing and environment: an unfamiliar room, sensors, a bedtime that differs from usual, or too little recorded sleep can change what appears in a single laboratory night.
This is why a sleep study report should be read as one integrated record. Stage percentages cannot replace the breathing, movement, arousal, oxygen, heart, or clinical findings.
When stage timing matters clinically
Stage information can help answer a focused diagnostic question, but it does not create a diagnosis on its own.
In suspected narcolepsy, clinicians may use an overnight polysomnogram followed by a standardized multiple sleep latency test. The test evaluates how quickly sleep begins during daytime nap opportunities and whether REM begins unusually soon. Adequate prior sleep, medication effects, substances, untreated sleep disorders, and test protocol all affect interpretation 11. A high or low REM percentage from one night or a wearable is not a narcolepsy test.
Some parasomnias are associated with particular states. Disorders of arousal such as sleepwalking and sleep terrors usually emerge from NREM sleep, often N3, but their diagnosis rests on the event history and appropriate differential evaluation, not the amount of N3 12. REM sleep behavior disorder involves dream-enactment behavior and loss of normal REM muscle atonia. It is not defined by having too much REM, and video polysomnography may be needed to document the relevant behavior and muscle activity 13.
What wearable sleep stages can tell you
Most watches, rings, and bedside trackers do not record the full EEG, EOG, and chin EMG montage used for clinical staging. They infer stages from movement, pulse signals, breathing, temperature, or combinations of indirect measurements. Some devices also combine N1 and N2 into labels such as “light” or “core” sleep, so their categories are not identical to a laboratory report.
A 2024 systematic review of three recent wrist-worn devices found moderate stage accuracy for two devices, but only eight eligible validation studies and continued room for improvement in stage-specific assessment 14. Performance also depends on the device, algorithm version, population, and whether sleep is disrupted.
Consumer data can still help a person notice broad patterns, such as a change in estimated sleep timing or duration. Compare trends within the same device and treat the stage graph as an estimate. The AASM states that consumer sleep technology should not be used to diagnose or treat a sleep disorder in place of appropriate clinical evaluation 15.
What to do about a low deep or REM sleep result
Do not try to force a stage with a supplement, soundtrack, temperature target, or a bedtime calculated from 90-minute blocks. First ask whether the number came from a clinical study or a consumer estimate, whether total sleep was sufficient, and whether the person has symptoms.
Discuss a clinical report with the clinician who interpreted it, especially if the recording was short, a medicine changed, or the report also shows breathing events, frequent arousals, unusual movements, or another abnormality. Seek an evaluation for persistent excessive daytime sleepiness, loud snoring or witnessed breathing pauses, sudden muscle weakness with emotion, repeated sleep attacks, injurious sleepwalking, or dream-enactment behavior.
The useful goal is adequate, well-timed, reasonably continuous sleep and treatment of any underlying disorder. A perfect-looking stage pie chart is neither required nor proof that sleep is healthy.
