Afternoon naps and sleep debt: what the experiments timed

Two laboratory studies, both from the same field, both measuring the minutes after a nap rather than the nap itself: ten minutes bought benefits lasting up to 155 minutes, thirty minutes bought a period of impaired alertness first.

goodWell supported, with the usual caveats about effect size.Revised 21 Oct 2026

A plain armchair with a folded pale blue cotton blanket over its back beside a small side table with a glass of water and a book with a blank cover

Both studies counted the same clock: the minutes after you sit up. That is where the payoff and the penalty live.

The nap is the most popular intervention in sleep science and one of the least examined, mostly because it is hard to blind. The two studies below are the closest thing the field has to a clean answer, and they both did something the popular articles about naps usually skip: they started their clock when the person sat up.

The first: twenty-four people, five nap lengths, three hours of testing

Brooks and Lack, in Sleep 2006, used a repeated-measures design with five conditions — a no-nap control, and naps of precisely 5, 10, 20 and 30 minutes of sleep. Nocturnal sleep was restricted to about five hours in participants’ homes, followed by afternoon naps at 3:00 PM and three hours of post-nap testing in a controlled laboratory. Twenty-four healthy young adults, good sleepers and not regular nappers.

The duration ladder, as measured:

5 minutes: few benefits compared with no nap at all.
10 minutes: immediate improvement in every outcome measure — sleep latency, subjective sleepiness, fatigue, vigor, cognitive performance — with some benefits maintained for as long as 155 minutes.
20 minutes: improvements emerging 35 minutes after napping, lasting up to 125 minutes.
30 minutes: a period of impaired alertness and performance immediately after napping, indicative of sleep inertia, followed by improvements lasting up to 155 minutes.

Their conclusion is that the 10-minute nap was overall the most effective of the lengths examined, and they suggest a process occurring in the first ten minutes of sleep that may account for the benefits.

The most counter-intuitive result is the 5-minute nap. Five minutes is not a nap, it is a lie-down, and the study says so with a specific verb: few.

The second: the one nobody expects

Hilditch, Centofanti, Dorrian and Banks, in Sleep 2016, ran the version that should be on the shelf of anyone who has ever felt worse after a nap. Thirty-one healthy adults aged 21 to 35, 18 of them female, in a three-day laboratory study with one baseline sleep from 22:00 to 07:00 and one experimental night randomised to total sleep deprivation, a 10-minute nap or a 30-minute nap, with nap opportunities ending at 04:00. Psychomotor vigilance, a digit-symbol substitution task, fatigue and sleepiness scales and self-rated performance were measured before the nap and at 2, 17, 32 and 47 minutes after it.

The mechanism first: the 30-minute nap contained 14.7 ± 5.7 minutes of slow wave sleep, the 10-minute nap contained 0.8 ± 1.5 minutes (P < 0.001). That is the whole phenomenon in one line. Ten minutes of clock time is not ten minutes of sleep; thirty is.

The cost: after the 30-minute nap, psychomotor vigilance performance deteriorated immediately from the pre-nap level and was still worse at 47 minutes post-nap (P < 0.015). In the no-nap condition it was also worse at 47 minutes. In the 10-minute nap condition there was no change across time at all.

And then the finding that should change how you use naps: all conditions — including the 30-minute nap that made people measurably worse — self-rated performance as better than pre-nap at every post-nap test point (P < 0.001). The authors state it directly: self-rated performance did not reflect objective performance following a nap.

What it means in practice

A thirty-minute nap leaves you measurably more awake and slower, and you do not notice. Psychomotor vigilance ran at 4.3 ± 0.3 responses per second before the 30-minute nap and 4.0 ± 0.5 at 47 minutes after it (P < 0.015), while self-rated performance at every one of those points read better than before.

The practical limit of both studies is generalisation: a 3:00 PM laboratory nap after five hours of sleep is not a 4:00 PM desk nap after a red-eye flight. But the mechanism they isolate — the first ten minutes count, and the minutes after them are paid for — is a fact about sleep physiology rather than about culture.

The debt a nap covers and the one it does not

Worth recalling what a nap does not do, using the strongest randomised evidence available. In Buxton and colleagues’ 2010 trial in 20 healthy men, eight or more nights with ten hours in bed followed by seven nights with five in bed reduced insulin sensitivity measured by intravenous glucose tolerance test by 20 ± 24 per cent (P = 0.001) and by clamp by 11 ± 5.5 per cent (P < 0.04).

That is a measure in days, not minutes. A nap does not repay that debt, and neither nap study measured it.

What the data do not show

Non mostrano che il sonnellino da venti minuti sia sempre peggiore: lo studio del 2016 valuta due sole durate in un contesto di turno notturno simulato, con occasione di sonnellino che terminava alle 04:00. Non mostrano che i dieci minuti di sonno lento siano la causa dei benefici — quello è un meccanismo inferito da Brooks e Lack, che parlano esplicitamente di un processo «che può spiegare» i benefici. E il 5% di differenza nelle auto-valutazioni è la parte più importante di entrambi i paper: la sensazione di essersi raddormentati non è una misura di essere svegli.

Il sonnellino da trenta minuti è l’unico intervento del sonno che ti fa star peggio e ti fa sentire meglio. È anche l’unico per cui gli esseri umani non hanno bisogno di un ricercatore per accorgersene.