The 15-to-20 Minute Nap Ceiling
Nap long enough to bank spindles, short enough to never surface out of deep sleep.
- Difficulty
- Starter
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 91%
The input is a mid-day dip in learning capacity. The mechanism is sleep spindles, bursting electrical brain waves that occur during non-rapid eye movement sleep. Walker's team ran a study where people learned a set of information at midday and a second new set at 6pm, with half the subjects napping in between and half remaining awake. The groups separated: nappers had a restoration of learning ability, non-nappers had a deterioration. Crucially, the greater the number of spindles during the nap, the greater the refreshment and recuperation of learning ability. Those research naps were 90 minutes, but Walker is clear that is a protocol choice, not a recommendation. The human sleep cycle from non-REM to REM is about 90 minutes, so a 90-minute nap gives both REM and non-REM an equal opportunity to show their prowess in transacting a functional benefit. For actual use, the short and long of it is: nap for about 15 to 20 minutes and no longer. Beyond that you start to descend into the deeper stages of non-REM, so waking after 30 or 40 minutes means coming out of deep sleep and hitting sleep inertia, which feels like a sleep hangover and is even worse than before you took the nap. There is a floor too: once you get down to about 9, 8, or 7 minutes, there are no measurable brain or body benefits. The output is a 15-to-20 minute window that captures cognitive benefit while avoiding the inertia penalty.
Origin
Walker's lab ran a nap study comparing learning capacity at midday and 6pm with and without a nap in the intervening 6 hours, and found the restoration tracked the number of sleep spindles. The 15-to-20 minute practical ceiling comes from the sleep inertia literature on waking out of deep non-REM.
Core principles
- 01Learning capacity deteriorates across a waking day and a nap restores it, so a nap is a reset rather than a top-up.
- 02The restorative agent is the sleep spindle, a bursting electrical brain wave in non-rapid eye movement sleep; more spindles means more restoration.
- 03Nap past 20 minutes and you descend into deeper non-REM, so waking at 30 or 40 minutes surfaces you out of deep sleep.
- 04Waking out of deep sleep produces sleep inertia, a sleep hangover that can leave you worse off than before the nap.
- 05Below roughly 7 to 9 minutes there are no measurable brain or body benefits, so there is a floor as well as a ceiling.
- 06The 90-minute nap used in research exists to give both REM and non-REM a fair chance to show a benefit, not because 90 minutes is the recommended dose.
How to run it
- 1
Nap on degradation, not on the clock
Take the nap when learning capacity has visibly dropped, which in Walker's study design was the middle of a 6-hour block between two learning sessions.
Pro tip Ferriss's parallel is that switching subjects also refreshes learning capacity, so a subject switch is the cheaper move if you cannot nap.
Watch out Napping on a fixed schedule regardless of state wastes the window and can interfere with night-time sleep pressure.
- 2
Set a hard alarm at 15 to 20 minutes
Time from lying down, not from falling asleep, and make the alarm non-negotiable. This is Walker's stated optimum: nap for about 15 to 20 minutes and no longer.
Pro tip Setting it from lying down builds in the sleep latency and keeps the actual sleep well inside the light non-REM stages.
Watch out A snooze-capable alarm converts a 20-minute nap into a 40-minute nap, which is the exact failure mode the ceiling exists to prevent.
- 3
Never wake at 30 or 40 minutes
If you overshoot 20 minutes you have started descending into deeper non-REM. Waking at 30 or 40 minutes surfaces you out of deep sleep and produces sleep inertia.
Pro tip If you have already overshot and can afford it, running the full 90-minute cycle is a better exit than waking at 40.
Watch out Walker says the sleep hangover from a 30-to-40 minute nap is even worse when you wake up than it was before you took the nap.
- 4
Do not bother below 9 minutes
Walker notes that once you get down to about 9, 8, or 7 minutes, they really do not see brain or body benefits. Sub-10-minute naps are not worth structuring your day around.
Pro tip If only 5 minutes is available, treat it as rest rather than expecting a cognitive return.
Watch out Assuming any sleep is better than none leads people to schedule micro-naps that measurably do nothing.
- 5
Reserve 90 minutes for full-cycle recovery only
The 90-minute figure is the human non-REM to REM cycle length used in lab protocols to give both stages a fair chance to demonstrate a benefit. Use it only when you deliberately want a complete cycle.
Pro tip Walker explicitly says he would not recommend the 90-minute nap for most people, despite it being the research standard.
Watch out Copying the study protocol as a personal habit imports a research design constraint as if it were a health recommendation.
In the wild
Walker's team had a group of people learn a whole set of information at midday, then brought them back at 6pm to learn an additional new set, so they could compare learning capacity at time point one against time point two. Half the subjects napped during the intervening 6-hour period; the others remained awake.
→ The groups separated: nappers showed a restoration of learning ability while non-nappers showed a deterioration, and the greater the number of sleep spindles during the nap, the greater the refreshment and recuperation of learning ability.
Ferriss asked how long the naps were and Walker said 90 minutes, immediately adding that he would not recommend that for most people. The reason for the length is that the human sleep cycle from non-REM to REM is about 90 minutes, so the protocol gives the sleeping brain a chance to run both stages and lets researchers see which stage correlates with the measured benefit.
→ A clean separation between a research design choice and a personal recommendation, with the practical guidance landing at 15 to 20 minutes instead.
Common mistakes
Napping 30 to 40 minutes
This is the worst possible duration. You have descended into deeper non-REM but wake before completing the cycle, so you surface out of deep sleep into sleep inertia, which Walker describes as feeling worse when you wake up than before you took the nap.
Copying the 90-minute research nap as a habit
The 90 minutes exists so REM and non-REM each get an equal opportunity to show a functional benefit inside a study design. Walker says directly he would not recommend that duration for most people.
Relying on very short micro-naps
Once you get down to about 9, 8, or 7 minutes, Walker says they really do not see brain or body benefits. A 5-minute nap does not deliver the spindle load that drives the restoration of learning capacity.
Is it for you?
Best for
Anyone running a long day of learning or cognitively demanding work who can carve out a genuine 20-minute window.
Not ideal for
People with insomnia running bedtime restriction therapy, where daytime napping bleeds off the adenosine the protocol depends on accumulating.
From the transcript
“the greater the number of spindles during the nap these sleep spindles that happen during what we call nonrapid eye movement sleep the greater the…”
“the short and long of it is nap for about 15 to 20 minutes and no longer”
“when you wake up after 30 or 40 minutes you're coming out of the deeper stage of sleep and you have what's called Sleep inera…”
“once you get to sort of down to about 9 8 or 7 minutes we really don't see brain or Body Benefits”
From the episode
#654: Dr. Matthew Walker, All Things Sleep Continued — The Hidden Dangers of Melatonin, Tools for Insomnia, Enhancing Learning and Sleep Spindles, The Upsides of Sleep Divorce, How Sleep Impacts Sex (and Vice Versa), Adventures in Lucid Dreaming, The One Clock to Rule Them All, The IP Addresses of Your Memories, and More