How much deep sleep do you need

Deep sleep is stage N3 of non-REM sleep, also called slow-wave sleep for the high-amplitude delta waves (0.5-2 Hz) that dominate its EEG. It carries the highest arousal threshold of any stage, which is why a person woken from N3 is groggy and disoriented (sleep inertia). Functionally it is the stage most associated with declarative-memory consolidation and, per Xie et al. 2013 (Science), glymphatic clearance of metabolic waste. It is distinct from REM sleep, which has a fast awake-like EEG and supports emotional and procedural processing; the two do different jobs and peak at opposite ends of the night.

The most-cited normative dataset is Ohayon et al. 2004, a meta-analysis of 65 polysomnography studies. In healthy adults, N3 sits at roughly 13-23% of total sleep time, which is about 60-110 minutes on a 7-8 hour night. There is no single 'correct' number to hit, and the percentage is age-dependent: it is highest in children and adolescents and declines substantially across adulthood. The practical reading is that total sleep time and a consistent early bedtime determine your N3 total far more than any nightly intervention.

N3 is not evenly spread. Slow-wave pressure, the homeostatic drive that builds with time awake, is highest at sleep onset and dissipates with each cycle (Dijk 2009). So the first one or two 90-minute cycles contain most of the night's N3, and the later cycles shift toward REM. This is the mirror image of REM, which is back-loaded toward morning. The practical consequence is that going to bed earlier adds mostly deep sleep, while sleeping in adds mostly REM, two facts worth separating when reading a hypnogram or a tracker graph.

The single largest change in sleep architecture across the lifespan is the decline of N3. Ohayon et al. 2004 quantified it: slow-wave sleep falls steeply from childhood through middle age, while REM stays relatively stable. By older adulthood, N3 can be a fraction of its young-adult level. This is normal aging, not a disorder, and it is why comparing your tracker's deep-sleep minutes to a younger person's, or to a fixed 'ideal,' is misleading. The age-adjusted baseline is the only honest comparison.

The lever that reliably increases N3 is increasing total sleep and protecting an early, consistent bedtime, because N3 rides the first cycles. Sleep deprivation produces a slow-wave rebound on recovery nights, evidence the brain homeostatically defends N3. Alcohol fragments sleep and can reduce later-night quality despite a sedative onset; many sleep medications increase light sleep more than N3. Intense exercise and a cool bedroom (~18 C) are plausibly supportive via core-temperature dynamics, but the effect sizes are modest. [VERIFY: claims that specific supplements or devices reliably increase N3 in healthy adults are largely small-sample or unsupported.]

Consumer wearables (Oura, Whoop, Apple Watch, Fitbit) do not record EEG. They infer sleep stages from heart rate, heart-rate variability, movement, and skin temperature, then apply a classifier. Stage-level output, especially N3, is the weakest part of these devices against polysomnography. [VERIFY: epoch-by-epoch N3 agreement vs PSG varies by device and firmware and is commonly well below overall sleep/wake agreement.] The useful signal is the trend over weeks, not the absolute minutes on any single night. If you want to influence the number, the only durable input is total sleep time and bedtime consistency. For the protocol, see the guide on how to get more deep sleep.