In short
The best tested nap is short: in the definitive duration trial, ten minutes delivered immediate alertness gains lasting over two hours with no grogginess, while thirty minutes bought the same benefit only after a period of impairment. The mechanism is sleep depth: past twenty to thirty minutes you enter deep sleep and wake into inertia. The window is the early afternoon dip, the alarm is not optional, and a late or long nap spends the sleep pressure the night needs. For insomnia, napping is counterproductive and the guidance says so.
The nap has a dose, like everything else in this Library. Most people take the wrong one.
01Naps work, sized correctly
The cleanest experiment in the field compared naps of five, ten, twenty and thirty minutes against no nap. Ten minutes won outright: immediate improvements in alertness and performance, sustained for as long as 155 minutes, with no post-nap grogginess. The twenty and thirty minute naps produced benefits too, but only after a period of impaired performance on waking.1 A nap is not a small night. It is its own intervention with its own dose response, and the dose is shorter than most people assume.
02Why longer backfires
The cost of the longer nap is sleep inertia, the transitional grogginess of waking, which runs fifteen to sixty minutes and is markedly worse when waking from deep sleep.2 Depth is a function of time: stay down past roughly twenty to thirty minutes and the brain descends into slow wave sleep, and an alarm at minute forty drags you out of the deepest part of the cycle. The ten minute nap works precisely because it ends before the descent begins.
03The protocol
Set an alarm for 15 to 20 minutes in bed, which yields roughly ten minutes of sleep after settling. Take it in the early afternoon, inside the circadian dip covered in the afternoon crash, where sleep comes easiest and the night is furthest away. Lying down in the dark without sleeping still counts as rest; do not fight for the nap. And end all napping by mid afternoon, because pressure spent late is pressure the night does not get back.
04The caffeine nap
The strangest trick in the literature is real: caffeine immediately before a short nap. Caffeine needs around twenty minutes to act, which is exactly the nap's length, so you wake as it arrives, stacking the nap's pressure relief with the stimulant's onset. In driving simulator studies the combination beat either alone against sleepiness.3 It is an occasional tool for genuinely dangerous drowsiness, not a daily habit, and the caffeine still counts against the day's cutoff arithmetic in the timing protocol.
05The cost side
Two honest costs. First, mechanism: a nap dissipates homeostatic sleep pressure, the currency the night runs on, per the two process model; long or late naps make the night lighter and later.4 Second, association: habitual napping beyond roughly 40 to 60 minutes tracks with worse metabolic outcomes in meta-analysis. That finding is observational and probably reflects reverse causation as much as harm, long nappers often nap because something is already wrong, and we grade it accordingly rather than waving it as a scare.5
06Who should not nap
The honest limit
If nights are the problem, naps are not the answer. For chronic insomnia, clinical guidance runs the other way entirely: consolidate sleep pressure into the night, which means no daytime napping at all while the night is being rebuilt.6 And a sudden new need to nap daily, in someone who never did, is worth a medical conversation rather than a protocol.
In the system
The nap slots into the same dip LAYER TWO is built to steady, and the pressure it spends is the physiology in sleep pressure. For the dip itself, see reducing the afternoon crash. Build your protocol →
Sources
Next in this thread
Reducing the afternoon crashThe dip the nap fits inside, and the rest of the order of operations.→How we handle evidence. The durations and timings above come from the cited trials; the metabolic association is graded as the observational finding it is, and the insomnia carve-out follows clinical guidance. Educational content, not medical advice.