Deep sleep: what it is, how much is normal and how far to trust the number
Deep sleep is the stage people most want more of. It is also the one wearables estimate least consistently. Here is the lab data and the device data side by side.
Deep sleep is the stage sleep researchers call N3 or slow-wave sleep. On an EEG it shows up as slow, large delta waves. It is hardest to wake someone from, it is concentrated in the first few hours of the night, and it responds to how long you have been awake. It is the stage sleep apps tend to celebrate. It is also the one where device numbers are least dependable.
How much deep sleep is normal
There is no single target, because deep sleep varies with age, sex and the person. Some reference points from polysomnography:
| Study | Year, journal | Who and how many | Design | Main result |
|---|---|---|---|---|
| Chinoy et al. | 2021, Sleep | 34 healthy adults, mean age 28 | Three lab nights with PSG | Average deep sleep 63 to 81 minutes depending on device subset, about 15 to 19% of sleep |
| Ohayon et al. | 2004, Sleep | 65 studies, 3,577 people aged 5 to 102 | Meta-analysis | In adults the percentage of slow-wave sleep fell with age; sleep efficiency kept falling after 60 |
| Boulos et al. | 2019, Lancet Respiratory Medicine | 169 studies, 5,273 healthy adults | Meta-analysis, current AASM scoring | Per decade of age, change in N3 percentage was -0.1% and not statistically significant; total sleep fell 10.1 min per decade |
| Redline et al. | 2004, Archives of Internal Medicine | 2,685 adults aged 37 to 92 | Community cohort, home PSG | Women had on average 106% more slow-wave sleep than men; age was linked to lighter sleep in men |
The two meta-analyses do not fully agree on age. The 2004 analysis found a clear fall in slow-wave sleep percentage in adults. The 2019 analysis, using newer scoring rules, found the decade-by-decade change in N3 percentage too small to be significant. Part of the difference is scoring: rules changed in 2007, and how strictly studies screened out people with health problems changed the age effect, as the 2004 authors pointed out. The direction most people experience, lighter and more broken sleep with age, holds in both; the bigger, consistent changes with age are in total sleep and time awake.
What changes deep sleep
- Alcohol. A 2013 review of every study on alcohol and normal sleep found slow-wave sleep increased in the first half of the night at all doses in most studies, while sleep in the second half became more disrupted. More deep sleep on a tracker after drinking is not a sign of better sleep.
- Sex. Women have more slow-wave sleep on average than men, by a large margin in the Sleep Heart Health Study.
- Sleep disorders. Untreated sleep apnoea fragments sleep. The Sleep Heart Health Study found sleep-disordered breathing was linked to poorer sleep, though the effect was smaller than age, sex and ethnicity.
- Some medicines and substances alter sleep architecture. If you suspect one, log it and compare weeks rather than nights.
How wearables estimate deep sleep
A ring or watch cannot see delta waves. It infers deep sleep from low movement, a slow and regular heart rate, higher HRV and, on some devices, temperature and the time of night. That works on average and misses on individual nights.
| Device | Deep sleep bias vs PSG, minutes, mean (SD) |
|---|---|
| Garmin Forerunner 245 | -52.4 (64.2) |
| Polar Vantage V | -40.4 (41.7) |
| WHOOP 3.0 | -19.6 (34.3) |
| Somfit | -11.7 (32.2) |
| Oura Ring Gen 2 | +2.4 (56.4) |
| Apple Watch S6 | Not reported separately (combined deep and REM) |
Read the standard deviations, not just the averages. Oura's average bias for deep sleep was close to zero in that study, but individual nights were off by an SD of 56 minutes. An early-generation Oura underestimated N3 by about 20 minutes in de Zambotti's 2019 study. In Chinoy's 2021 study, the two Garmin models and Fitbit had deep sleep biases that were not statistically significant on average. The 2024 Oura-funded validation of its Gen3 ring with a newer algorithm found no significant difference from home PSG for time in deep sleep across 96 adults. Devices and software change often, so a validation of last year's model is only a guide to this year's.
Should you chase more deep sleep?
Not directly. There is no evidence that pushing a tracker's deep sleep number up improves health, and the measurement error on any one night is larger than most of the changes people try to make. What does have evidence is the boring list: enough total sleep, a regular schedule, less alcohol in the evening, and treatment for apnoea if you snore heavily and wake unrefreshed. The sleep and health guide has the trial data, and peptides for sleep covers what the research does and does not show for compounds marketed for sleep.
Reading your own trend
If deep sleep drops for a week on the same device, look first at timing and alcohol, then at anything new in training, stress or medicines. A single night with 30 minutes rather than 70 is within normal error. A month-long shift is worth more attention. See reading wearable trends.
A note on older devices
Most of the independent data above come from devices that are now one to three generations old. That is the normal lag between a product launch and a published validation. When a new model claims better staging, look for an independent study before treating the claim as settled.
Logging it
Peptide AI pulls deep sleep minutes from your ring or watch through wearable integrations and draws them on the same axis as the dose log and evening meals. That makes it easier to see whether a change in deep sleep follows something you did or is just a noisy night.
Questions people ask
How much deep sleep do I need?
There is no fixed target. In a 2021 lab study of healthy adults averaging 28 years, PSG recorded roughly 63 to 81 minutes of deep sleep a night, about 15 to 19 percent of total sleep. Amounts vary with age and sex.
Does deep sleep decrease with age?
A 2004 meta-analysis of 3,577 people found slow-wave sleep percentage fell with age in adults. A 2019 meta-analysis of 5,273 adults using newer scoring found the per-decade change in N3 percentage was not significant, while total sleep fell 10.1 minutes per decade.
Why did my deep sleep go up after drinking?
Alcohol increases slow-wave sleep in the first half of the night in most studies, then disrupts sleep in the second half. It is not a sign of better sleep.
How accurate is deep sleep on Oura, WHOOP or Garmin?
It varies. In a 2022 study of 53 adults, average deep sleep bias ranged from +2.4 minutes (Oura Gen 2) to -52.4 minutes (Garmin Forerunner 245), with individual-night standard deviations of 34 to 64 minutes.
Sources
- Performance of seven consumer sleep-tracking devices compared with polysomnography (Chinoy et al.), Sleep, 2021
- Meta-analysis of quantitative sleep parameters from childhood to old age in healthy individuals (Ohayon et al.), Sleep, 2004
- Normal polysomnography parameters in healthy adults: a systematic review and meta-analysis (Boulos et al.), Lancet Respiratory Medicine, 2019
- The effects of age, sex, ethnicity, and sleep-disordered breathing on sleep architecture (Redline et al.), Archives of Internal Medicine, 2004
- Alcohol and sleep I: effects on normal sleep (Ebrahim et al.), Alcoholism: Clinical and Experimental Research, 2013
- A validation of six wearable devices for estimating sleep, heart rate and heart rate variability in healthy adults (Miller et al.), Sensors, 2022
- The sleep of the ring: comparison of the OURA sleep tracker against polysomnography (de Zambotti et al.), Behavioral Sleep Medicine, 2019
- Validity and reliability of the Oura Ring Generation 3 with Oura sleep staging algorithm 2.0 compared with multi-night ambulatory polysomnography (Svensson et al.), Sleep Medicine, 2024
