Where the 20-minute nap rule actually came from

What everyone says

The standard advice comes in one of two shapes. Either "nap for 20 minutes, any longer and you will wake up groggy," or the fuller version: "nap for 20 minutes or 90 minutes, nothing in between." Both are usually delivered with a NASA reference attached, and often with the specific figures 34% and 54% quoted alongside.

I want to be clear before pulling this apart: the advice is broadly reasonable. Short naps do help, and the mechanism behind the ceiling is real. What is wrong is the certainty and the sourcing. Almost nobody repeating "20 minutes" can tell you which study produced that number, and when you go looking, the honest answer is that no study did.

Where 20 minutes actually comes from

This one surprised me, because unlike the 10-3-2-1-0 rule, the trail does not end at a blog post. It ends somewhere respectable.

The National Institute for Occupational Safety and Health, part of the CDC, publishes fatigue-countermeasure training for nurses working long hours. Its napping module recommends a daytime brief nap of less than 20 minutes, and gives the two-window version explicitly: waking either at around 20 minutes, before going into deeper sleep, or at around 90 minutes, at the predicted end of a sleep cycle, should mean less grogginess from sleep inertia. It notes that both short naps of 15 to 30 minutes and longer ones of around 90 minutes can raise alertness, that a brief nap helps for a couple of hours, and (the part most articles drop) that naps are a temporary aid to alertness and not a substitute for adequate regular sleep. The same module records that the American Academy of Sleep Medicine's Standards of Practice Committee recommends planned naps before and during night shifts for people struggling with shift work.

Worth noticing that NIOSH says less than twenty, which points in the same direction as the study below rather than at twenty itself.

So there is a genuine American institutional source for the 20-minute figure. That is more than most viral sleep numbers can claim.

But look carefully at how NIOSH frames it. Twenty minutes is offered as the point before you descend into deep sleep, a ceiling, reasoned from the architecture of sleep. It is not presented as the duration that produced the best results in a trial, because that is not what it is. The number answers the question "how long can I nap without waking up groggy?" It does not answer "how long should I nap?" Those are different questions, and the advice has quietly swapped one for the other.

What the NASA study really found

The study attached to almost every power-nap article is real, and it is worth reading properly rather than through the summary of a summary.

Mark Rosekind and colleagues at the NASA Ames Fatigue Countermeasures Program studied 21 volunteer pilots flying scheduled overnight trans-Pacific Boeing 747 routes. Twelve were given a planned 40-minute rest opportunity during low-workload cruise; nine continued normal duties. Inside that window, the rest group took about 5.6 minutes to fall asleep and slept a mean of roughly 25.8 minutes.

The results the report actually prints are counts, and they are more striking than the version usually quoted. Across the final 90 minutes of flight, descent and landing, the part that matters, the rested pilots recorded 34 microsleep events against 120 in the group that had not rested. On psychomotor vigilance testing, the rest condition produced 81 lapses against 124.

Why this does not support "20 minutes"

Two reasons, and they compound.

First, the protocol was a 40-minute scheduled window, and 25.8 minutes is what the pilots happened to sleep inside it. That is a measurement, not a prescription. Nobody chose 26 minutes as a dose and nobody tested it against alternatives. Rounding a measured average down to "20 minutes" and presenting it as the recommended nap length inverts what the study did.

Second, the population was commercial pilots partway through an overnight long-haul flight, already accumulating fatigue on a badly disrupted schedule. The effect of a rest period on someone in that state tells you little about a well-slept office worker at 2pm. It is a fatigue-countermeasure study, not a productivity-hack study.

The study that compared nap lengths

Here is the paper that should be cited in every one of these articles and almost never is.

Amber Brooks and Leon Lack, at Flinders University in Australia, published A Brief Afternoon Nap Following Nocturnal Sleep Restriction: Which Nap Duration Is Most Recuperative? in SLEEP in 2006. It is not American, and I have leaned American elsewhere on this site, but on this specific question it is the study that exists.

The design is better than I first gave it credit for. Twenty-four healthy young adults, twelve men, twelve women, mean age 22.5, all good sleepers, none habitual nappers, all low caffeine consumers. Crucially it is within-subject: every participant went through all five conditions rather than being sorted into five groups, with the order counterbalanced and at least two normal nights between sessions. That removes a great deal of the between-person variation that makes small sleep studies hard to read.

Each session began with the night before cut to five hours, from 2am to 7am. Participants came to the laboratory at 1pm and napped at around 3pm, then were tested repeatedly for three hours afterwards, at 5, 35, 95 and 155 minutes after waking, with additional sleep-latency measurements at 65, 125 and 185 minutes. The nap conditions were precisely 5, 10, 20 or 30 minutes of actual sleep, not time spent lying down, which is a distinction most reporting of this study loses.

The authors' conclusion was that the 10-minute nap was the most effective of the four durations tested.

The bit worth sitting with

Look at the 20-minute row again. The benefits did not appear for 35 minutes after waking. If you are napping at your desk on a 20-minute timer because you have a meeting shortly, the study says you will attend that meeting during the dead zone, and the benefit will arrive after you no longer need it.

That is the opposite of how the advice is usually given. "Twenty minutes" is sold as the efficient option, short, sharp, back to work. On this evidence, ten is the efficient option, and twenty carries a delay nobody mentions.

I want to be fair to the limits. Twenty-four people is a small sample, they were all young adults with good sleep habits, and every session followed a night deliberately cut to five hours, so this tells you about napping while sleep-deprived, which is not the same as napping on a normal day. It is one laboratory, and I have not surveyed whether it has been replicated, so I am not going to imply either that it has or that it has not.

What I will say is that the within-subject design makes it considerably stronger than a small sample usually implies, and that it is the closest thing to a direct comparison of nap durations I could find. When the only head-to-head test points away from the number everyone repeats, that is worth knowing even if it does not settle the question.

Sleep inertia, the thing the rule is avoiding

The mechanism behind the 20-minute ceiling is sound, and it is the reason I would not throw the advice out.

Slow-wave sleep, stage N3, the deep stuff, typically does not begin until roughly 30 minutes into a nap. Wake someone out of N3 and they get sleep inertia: a stretch of genuinely impaired alertness and cognition that can take 15 to 30 minutes to clear. This is not a vague feeling of grogginess. It shows up on reaction-time testing.

So the logic runs: stay under about 30 minutes, and you should still be in lighter N1 and N2 sleep when the alarm goes, and wake cleanly. Twenty is a conservative margin inside that 30-minute boundary.

That reasoning is good. It is also, precisely, reasoning, a ceiling derived from sleep architecture rather than an optimum found by experiment. And the 30-minute row in Brooks and Lack shows the boundary is real: 30 minutes was the one duration that produced measurable impairment immediately after waking.

Which leaves 20 minutes in an odd position. It is safely under the inertia threshold, but the one study that checked found it slower to help than a nap half its length. The ceiling is well founded. The number sitting just beneath it is not the best answer to a question the ceiling was never asked.

The 90-minute nap

The other half of the advice, "20 minutes or 90, nothing in between", has better support than people realise, and it comes from a different place entirely.

Sara Mednick, Ken Nakayama and Robert Stickgold published Sleep-dependent learning: a nap is as good as a night in Nature Neuroscience in 2003, out of Harvard. They found that naps of 60 to 90 minutes containing both slow-wave sleep and REM produced improvement on a visual texture-discrimination task comparable in magnitude to a full night of sleep, and that the benefit was additive with the following night's sleep.

That is a real and quite striking finding. Note what it is about, though: it is about memory consolidation and perceptual learning, not about feeling alert this afternoon. The 90-minute nap and the 10-minute nap are not competing options for the same goal. One is for locking in something you have just learned. The other is for not falling asleep in a meeting. Articles that present them as two settings on the same dial have flattened two different literatures into one tip.

The 90 also carries the same caveat as everywhere else on this site: cycle length averages 90 minutes but ranges from roughly 70 to 120, so a 90-minute alarm does not reliably land at the end of a cycle. It is a decent bet, not a guarantee.

So how long should you nap?

My reading of the evidence, stated as what it is, a reading, not a guideline:

• If you need to be sharp soon: 10 minutes. It is the only duration shown to help immediately on waking, and some of its benefits were still measurable when that study stopped testing at 155 minutes. It is also the easiest to actually fit into a day.
• If you have the time and no meeting to get to: 20 minutes is fine, but expect the benefit to arrive later than the alarm does. Do not schedule anything demanding for the half hour after.
• Avoid the 25 to 45 minute band if you can. That is where you risk waking out of slow-wave sleep and paying for it with inertia.
• If you are consolidating something you have just learned: 60 to 90 minutes. Different goal, different literature, and the evidence for it is good.
• If you are chronically short on sleep, none of this is the fix. Naps are a countermeasure. The NASA pilots needed one because their schedule was brutal, not because napping is a technique for optimising an otherwise adequate night.

And if you take one thing from this page, take the thing that has nothing to do with duration: almost every confident number in sleep advice is a ceiling, a midpoint or a rounded measurement wearing the costume of a finding. Twenty minutes is a reasonable ceiling. It was never a result.

What this page does not cover

This is about healthy adults choosing a nap length. It is not about narcolepsy, shift-work disorder, or the daytime sleepiness that comes with untreated sleep apnoea. If you are falling asleep involuntarily during the day, or need naps to get through a normal day, that is worth a doctor's time and no timer will fix it.

I have also not tested any of this on myself in a structured way, and I am not going to imply otherwise. Everything above is a reading of published sources.

A self-experiment is tempting, and I want to be honest about how little it would prove. Brooks and Lack counterbalanced five conditions across 24 people and put at least two normal nights between sessions precisely because a one-person alternating-day version is swamped by confounders: the previous night's sleep, caffeine, time of day, accumulated sleep debt, order effects, and getting better at the reaction-time task simply through repetition. If I run one it will appear here labelled as what it is, a personal observation, not evidence about anybody else.

Questions people actually ask

References

• Brooks A, Lack L. A Brief Afternoon Nap Following Nocturnal Sleep Restriction: Which Nap Duration Is Most Recuperative? SLEEP. 2006;29(6):831–840. Within-subject comparison of no-nap, 5, 10, 20 and 30-minute naps in 24 healthy young adults (12 male, 12 female, mean age 22.5) after nights restricted to five hours. Post-nap testing at 5, 35, 95 and 155 minutes, with sleep-latency measures at 65, 125 and 185 minutes. Source for the 10-minute finding, the 35-minute delay on the 20-minute nap, and the sleep inertia seen at 30 minutes. Read in full and checked against this page.
• Rosekind MR, Graeber RC, Dinges DF, Connell LJ, et al. Crew Factors in Flight Operations IX: Effects of Planned Cockpit Rest on Crew Performance and Alertness in Long-Haul Operations. NASA Technical Memorandum, held by the US DOT National Transportation Library. Source for the 40-minute rest protocol, the 5.6-minute sleep latency, the ~25.8 minutes of actual sleep, and the microsleep and vigilance-lapse counts. Not the printed source of the 34% / 54% figures.
• Mednick S, Nakayama K, Stickgold R. Sleep-dependent learning: a nap is as good as a night. Nature Neuroscience. 2003;6(7):697–698. Source for the 60 to 90-minute nap and perceptual learning.
• National Institute for Occupational Safety and Health (CDC). Training for Nurses on Shift Work and Long Work Hours, Module 7: Napping, an Important Fatigue Countermeasure. Source for the 20-minute and 90-minute waking windows and the AASM Standards of Practice position on planned naps for shift work.
• Reviews of sleep inertia and the timing of slow-wave sleep onset, for the roughly 30-minute delay before N3 and the 15 to 30-minute recovery from waking out of it.

Related reading

• Where sleep numbers come from, the source index: the 20-minute nap alongside seven other sleep numbers traced to their origins.
• The 10-3-2-1-0 sleep rule, a viral countdown traced back to the productivity author who invented it.
• How long a sleep cycle actually is, why the 90 in "90-minute nap" is a midpoint of 70 to 120.
• How many hours of sleep is 11pm to 7am?, for the night rather than the afternoon.