How long should a nap be? For most of my twenties I would have said 45 minutes, and I would have said it with confidence, because 45 sounded moderate. Not a lazy nap, not a useless one. I took exactly that nap before a night shift at the call center one winter. I set the alarm, woke on time, and felt like my skull had been packed with wet sand. In the first hour I read two account numbers back to callers wrong and had to apologize to a man for getting his own address wrong.
That nap went into the spreadsheet, and so did the ones after it. The rows showed something the research had worked out long before I started counting. A nap has two good lengths and a bad middle: around 20 minutes, or around 90. Most people who wake somewhere between 30 and 60 minutes feel worse than before they lay down. Here is why, and how I pick between the two.
Why 30 to 60 minutes is the worst place to wake up
Once you drift off, you go through light sleep, then deeper sleep, then back up into a lighter stretch and dreaming sleep. The whole loop takes roughly 90 minutes. Carskadon and Dement put the cycle there in Principles and Practice of Sleep Medicine, with plenty of room either side, and yours may be 70 or 110. I wrote more about that loop in the piece on the 90 minute sleep cycle.
For a nap, the deep stretch is the part that counts. Sleep researchers call it slow wave sleep. In plain words, it is the stage where your brain waves go slow and large and you are hardest to wake. In a rested adult it usually starts after the first 20 to 30 minutes of sleep. That is a rule of thumb, and it moves. After a bad night you can sink into deep sleep faster.
If you wake someone out of deep sleep, they get sleep inertia, the heavy, slow feeling right after waking. Tassi and Muzet reviewed the research on it in Sleep Medicine Reviews in 2000, and the pattern held up. Inertia is worse when you are pulled out of deep sleep, and it can last anywhere from a few minutes to half an hour or more. I covered the fog itself in why you feel groggy after waking up.
Put those two facts together and the trough explains itself. A 45 minute nap has a good chance of ending in the middle of your deepest sleep. You get the grogginess of a long nap without the full cycle that would have carried you back out of it. That was my winter afternoon.
The 20 minute nap, and what the studies found
The short nap works because it ends before deep sleep has a real grip. You get light sleep, some of the pressure to sleep eases off, and you wake from a stage that lets go of you easily.
The study I trust most here is Brooks and Lack, published in Sleep in 2006. They cut a group of young adults down to about five hours of sleep the night before. The next afternoon each person got a nap of 5, 10, 20 or 30 minutes, or no nap at all, and was then tested for alertness and performance. The 5 minute nap did little. The 10 minute nap gave the quickest lift, with benefits showing almost straight away and lasting a couple of hours. The 20 minute nap helped too, though the benefit took a little longer to arrive. The 30 minute nap brought a stretch of grogginess first and only then an improvement.
Know a few things before you copy that. It was a small lab study of young, healthy sleepers. The naps were counted as minutes of actual sleep, not minutes in bed. And everyone was already short on sleep, which is when most people take a nap anyway, so I think it transfers to real life better than most lab work.
The other study people quote is NASA's work with long haul pilots, led by Mark Rosekind in the 1990s. Pilots were given a 40 minute rest window in the cockpit seat during a flight and slept about 26 minutes of it on average. The pilots who napped had fewer lapses on reaction time tests late in the flight than the ones who did not. That is where the "NASA nap" you see everywhere comes from. It is a fair result, though a nap in a quiet cockpit is not quite your couch.
So the best nap length for a quick power nap is around 10 to 20 minutes of actual sleep. I aim for about 15. I rarely drop off the moment I lie down, so that means an alarm 20 to 25 minutes after my head hits the pillow.
The 90 minute nap, when you have the afternoon
The other good nap length is a full cycle. Sleep for roughly 90 minutes and you go down through deep sleep, come back up through the lighter stages and dreaming sleep, and wake near the end of the loop, where waking is easy again.
I take this nap on days when I am properly behind. It also does something the short nap does not. In Nature Neuroscience in 2003, Mednick and colleagues found that a 60 to 90 minute nap containing both deep sleep and dreaming sleep improved a visual learning task about as much as a full night of sleep did. That was one task in one lab, and it says nothing about your work inbox. Still, it is a real difference between a long nap and a short one.
The costs are real too. You need a clear two hours once you count falling asleep and getting going again. And if your cycle runs shorter or longer than 90 minutes, you can still land in a bad spot, just later. When a 90 minute nap leaves me groggy, I move the alarm by 10 minutes the next time and write down what happened.
There is a middle option I should be straight about, because the nap calculator offers it: a 60 minute recovery nap. It sits in the stretch I warned you off earlier, and it will probably end in deep sleep. I take it only on days I am so far behind that I need the deep sleep more than a clear head, and I cannot find two hours. Expect to wake groggy, and give yourself half an hour before anything that needs you sharp, driving included. If the extra 30 minutes are there, take the 90.
A long nap also uses up some of the sleep pressure you need for tonight. If you nap long most days because your nights are short, add that up properly. The sleep debt calculator shows how big the hole is, and my notes on sleep debt cover why a nap only pays back part of it.
Count the minutes it takes to fall asleep
I made this mistake for years, and it is why "set an alarm for 20 minutes" so often goes wrong. The nap length that matters is how long you are asleep, not how long you are lying down.
The gap has a name: sleep latency, which means how long it takes you to fall asleep once you lie down meaning to sleep. For rested people at night it is often around 10 to 20 minutes. In the afternoon it swings a lot. If you are very tired it might be three minutes. If you had a coffee at lunch it might be never.
So I add my usual latency to the sleep I want and set the alarm for the total.
- For a short nap, I want about 15 minutes of sleep plus my usual 5 to 10 minutes to drift off, so the alarm goes 20 to 25 minutes after I lie down.
- For a full cycle, I want about 90 minutes of sleep plus the same few minutes, so the alarm goes 95 to 100 minutes out.
That arithmetic is what the nap calculator is for, so you do not have to do it while you are already half asleep.
If you do not know your latency, guess 10 minutes and adjust. If you keep waking groggy from a short nap, you are probably falling asleep faster than you think, and the alarm should come earlier.
When in the day to nap
When you nap changes how it feels almost as much as how long you nap. Most people hit a natural dip in alertness in the early afternoon, often somewhere around 1 to 3 p.m. It comes from your body clock and not only from lunch, though lunch does not help. A nap inside that dip comes easier, because you fall asleep faster.
The later in the day you nap, the more it eats into tonight. Sleep pressure is the drive to sleep that builds the longer you are awake, and it is part of what gets you to sleep at bedtime. A nap spends some of it. My rule of thumb is no short nap after about 3 p.m. and no long nap after about 1 p.m. when I go to bed at a normal hour. That comes from my spreadsheet, not a study. If you work nights, move those times to match your own bedtime.
The coffee nap fits here too. Reyner and Horne, in Psychophysiology in 1997, put sleepy drivers in a simulator. Coffee drunk right before a nap of about 15 minutes cut driving incidents more than the coffee alone did. Caffeine takes roughly 30 to 45 minutes to peak in the blood, so it arrives about when the nap ends. I tried it for a month. On some days it worked. On others it left me jittery, so I save it for afternoons that matter. If you drink that coffee late in the day, run the time through the caffeine cut-off calculator first.
If you need a nap every day even after full nights in bed, or you fall asleep when you did not mean to, talk to a doctor, and you can read what I am and am not on my about page.
Setting the alarm for your next nap
My spreadsheet has a column for nap length and a column for how I felt an hour later, scored one to five. The 20 minute naps average a four. On days I had the time, the 90 minute naps average a four too. The 40 to 50 minute naps average a two. I kept taking them long after I should have known better, and a few are a one with a note that just says "sand".
So pick a length before you lie down, not after you wake up. Decide whether you have 25 minutes or two hours. If you have something in between, take the short nap and spend the rest of the time on something else, unless it is one of those far-behind days for the 60 minute nap.
Tonight, open the nap calculator and plan tomorrow's nap around your afternoon dip. Write the alarm time down. When you wake from the nap, give yourself a number from one to five. A week of those numbers will tell you your own best nap length better than any study can.