Sleep Calculator

What this calculator does differently

Most sleep calculators treat 90 minutes as an exact cycle and hand you an instruction: go to bed at 10:45 pm. That level of precision is not supported by measured data. This page does not hide the fact. It calculates the two things that survive the uncertainty instead — your total sleep, and what to do about the nights when you cannot get it.

Why 90-minute timing cannot be precise

The average is wrong before you even get to the spread. A 2024 analysis of 6,064 cycles from 369 polysomnography recordings put the median cycle length at 96 minutes (mean 99.5). A 2025 study using subcutaneous EEG measured 98 ± 29 minutes.

Six minutes is not the interesting part. The variability is. The coefficient of variation between one cycle and the next came to 29.3 per cent, and the measured standard deviation of a single cycle is about 29 minutes.

Here is what that does to the arithmetic. Uncertainty accumulates with the square root of the number of cycles, so chaining five of them carries 29 × √5 ≈ 65 minutes of error — more than half a cycle. "Go to bed at 11:15 pm and five cycles later you surface at 7:00" actually means "somewhere around 6:00, or maybe 8:00." At that width, aligning your alarm to a cycle boundary is not a thing anyone can do.

Cycles are not evenly spaced either. Measured directly, the first cycle runs 92 ± 29 minutes and the second 111 ± 29. Multiplying 90 by a whole number is off from the very first cycle.

And the drift is worst exactly when you would most want the calculator. Higher sleep pressure — that is, being short on sleep — lengthens the NREM portion of the first cycle. The night you most want to squeeze value out of a cycle calculation is the night it fits worst.

None of this is a fringe view. Australia's Sleep Health Foundation puts person-to-person cycle variation at roughly 60 to 110 minutes and says outright that the assumptions these calculators make are not based on scientific evidence.

So this page gives you a range, not a single minute. If the range looks uncomfortably wide, that width is the honest one.

What is certain instead — the total

Cycle boundaries are out of reach, but total sleep is exact arithmetic. It is also the number that the health guidance is written about.

The joint consensus of the American Academy of Sleep Medicine and the Sleep Research Society states that adults aged 18 to 60 should sleep 7 or more hours per night on a regular basis. The NHS gives the target as around 7 to 9 hours. This calculator uses 7 to 9 hours as the adult band, and both sources are explicit that the requirement varies between individuals.

Children and teenagers need more, which is why the calculator asks your age rather than assuming an adult. The bands used here are 9 to 12 hours for primary-school children and 8 to 10 hours for secondary-school students — the figures set out in Japan's Ministry of Health, Labour and Welfare sleep guide, which is the set of age bands this page has checked against a primary source. National bodies elsewhere publish their own figures and they differ in the detail, so read the band as the shape of the recommendation rather than as a threshold. One gap worth stating plainly: no primary source was confirmed for the under-5 band shown here (10 to 13 hours), so treat that one as indicative only. It will be replaced when a source is confirmed.

What all of this means in practice is simple. Giving up 40 minutes of sleep to land on a cycle boundary runs the wrong way — you are trading a certain quantity for an uncertain boundary.

Short nights — the nap is 10 minutes

This is the part the calculator most wants to be useful for. "Get seven hours" is no answer at all to someone who has four.

In an experiment where night sleep was restricted to 5 hours, participants took 5, 10, 20 and 30-minute naps at 3 pm. The results run against intuition.

• 10 minutes — every measure (sleep latency, subjective sleepiness, fatigue, vigour and cognitive performance) improved immediately on waking, and some of the benefit was still measurable 155 minutes later. Best of the four overall.
• 20 minutes — the benefit did not arrive until 35 minutes after waking, then ran for 125 minutes.
• 30 minutes — produced a period of reduced alertness and worse performance right after waking, with improvement only after that had passed.

A separate experiment tested 30-second and 90-second naps and found no measurable improvement over not napping at all. So there is a floor below which nothing happens, and above that floor longer does not mean better.

Why 30 minutes backfires — you reach deep sleep

The mechanism was measured, not guessed at. The 30-minute nap accumulated an average of 14.7 ± 5.7 minutes of slow-wave sleep, and 80 per cent of participants were woken out of it. The 10-minute nap accumulated 0.8 ± 1.5 minutes of slow-wave sleep, and 80 per cent woke from light stage 2 sleep.

The fogged, slow state you get when pulled out of deep sleep is called sleep inertia. To be straight about the limits of the evidence: the authors of that experiment noted that they could not isolate an independent effect of slow-wave sleep, and a 2019 review concluded that it is difficult to draw a clear conclusion about the role of deep sleep in sleep inertia. The direction is consistent across studies; the causal mechanism is not settled.

The same principle shows up in operational research. In a NASA study of long-haul pilots, 93 per cent of planned 40-minute rest opportunities produced actual sleep, averaging 26 minutes. The composition was 30 per cent stage 1, 62 per cent stage 2, 8 per cent slow-wave sleep and 0 per cent REM — they barely touched deep sleep. Reaction speed improved by 10 to 16 per cent, and lapses of attention numbered 81 in the rest group against 124 without.

The most important warning — do not trust how you feel

The single most useful finding in the nap experiment is this one. The people who took the 30-minute nap were measurably worse on objective testing, and rated themselves as better than before the nap. Not marginally, and not once: significantly so at all four post-nap measurement points (p<0.001).

In other words, subjective freshness and actual recovery moved in opposite directions. "I feel fine now" is not evidence that you have recovered. If you are about to drive, or to make a call that matters, manage the nap by the clock rather than by how it feels.

It is also why this calculator never promises a time at which you will "wake up refreshed." Refreshed is not something you are in a position to assess.

Age matters; sex barely does

How late you are inclined to sleep — your chronotype — moves with age, reliably. In a survey of roughly 25,000 people, the tendency to go to bed late increases through adolescence, peaks at around age 20, and advances steadily from then on for the rest of life. The authors proposed that turning point as a biological marker for the end of adolescence. The same curve was reproduced in 800 people in South Tyrol, whose way of life is nothing like that of the original sample.

The size of the effect is not small either: about 1.9 hours of movement in the ten years between 15 and 25. That is worth knowing before writing off a teenager's late nights as laziness.

Sex, by contrast, is too small to build an input around. In the same line of research the difference between men and women peaked at age 21.6 at 0.27 hours — 16 minutes — disappeared at 41.4, and then reversed sign, with women becoming the later group. A 16-minute maximum that flips direction in middle age would add nothing to a bedtime estimate. That is why this calculator does not ask.

For completeness: no primary source showing that required sleep duration itself differs by sex was confirmed in this round of checking. If one is, it will be reflected here.

Basis for the figures

• Median cycle 96 min, mean 99.5, between-cycle CV 29.3% (6,064 cycles, 369 recordings) — Cajochen et al., Sleep Health 2024;10:S52–S62
• Cycle 98 ± 29 min; first cycle 92 ± 29, second 111 ± 29 — Leguia et al., Brain Commun 2025;7(6):fcaf469
• 10-min nap best overall (immediate onset, 155 min), 20-min (onset after 35 min, 125 min), 30-min (sleep inertia) — Brooks A & Lack L, Sleep 2006;29(6):831–840
• Slow-wave sleep 14.7 ± 5.7 min in the 30-min nap (80% woken from SWS) vs 0.8 ± 1.5 min in the 10-min nap (80% woken from stage 2); subjective/objective divergence p<0.001 — same study
• 30-second and 90-second naps produced no improvement — Tietzel & Lack
• Pilots slept a mean 26 min in planned rests; reaction speed 10–16% better; 81 vs 124 lapses of attention — NASA TM-108839
• Chronotype latest at around age 20, advancing thereafter (N ≈ 25,000; replicated in 800 people in South Tyrol) — Roenneberg et al., Curr Biol 2004;14(24):R1038–9
• Sex difference max 0.27 h at age 21.6, gone at 41.4, sign reversed after — Fischer et al., PLoS One 2017
• Adults 18–60 should sleep 7 or more hours — AASM & Sleep Research Society consensus (Watson NF et al., J Clin Sleep Med 2015;11(6):591–592)
• Around 7 to 9 hours for a healthy adult — NHS
• 9–12 hours (primary school) and 8–10 hours (secondary school) — Japan MHLW sleep guide 2023; the under-5 band used here has no confirmed primary source
• Person-to-person cycle variation approx. 60–110 min; calculator assumptions not evidence-based — Sleep Health Foundation (Australia)
• Role of deep sleep in sleep inertia unresolved — Hilditch CJ & McHill AW, Nat Sci Sleep 2019;11:155–165

This is a scheduling tool, not medical advice. Persistent trouble falling asleep, waking repeatedly in the night, or being sleepy all day is worth raising with a doctor — those are treatable conditions, and no amount of bedtime arithmetic addresses them. The nap guidance here assumes a healthy adult dealing with a temporary shortfall.