General information, not medical advice. Persistent insomnia, daytime sleepiness or suspected sleep apnoea are worth a doctor’s appointment, not an internet article.
Eight hours is the most confidently repeated number in health. It is printed on mattresses, built into phone apps, and used by millions of people as a nightly pass-fail test.
It appears in no major sleep guideline.
What the guidelines actually say
The authoritative statement is the 2015 joint consensus of the American Academy of Sleep Medicine and the Sleep Research Society, produced by 15 panellists reviewing 5,314 articles under a formal RAND appropriateness process over twelve months.
Its recommendation, in full: adults should sleep seven or more hours per night on a regular basis to promote optimal health.
That is a floor, not a target. There is no upper figure. The statement explicitly declines to set one, noting that more than nine hours may be appropriate for young adults, people recovering from sleep debt and the ill — and that for everyone else, it is uncertain whether sleeping more than nine hours carries health risk.
The CDC repeats the same formulation: seven or more hours for adults aged 18 to 60.
Somewhere between “seven or more” and the bedside table, an open-ended minimum became a precise quota that people can fail every single night.
What people sleep without electric light
The obvious test is to find people living without artificial lighting and measure them. Gandhi Yetish and colleagues did exactly that, publishing in Current Biology in 2015: actigraphy on 94 individuals across the Hadza of Tanzania, the San of Namibia and the Tsimane of Bolivia, covering 1,165 recorded days.
Every group mean landed between 5.7 and 7.1 hours. Sleep efficiency — time asleep as a share of time in the sleep period — ran at 81 to 86%, meaning roughly an hour or more awake inside the night was routine even there.
Chronic insomnia, meanwhile, was almost absent: regular trouble falling asleep in 1.5% of people and trouble staying asleep in 2.5%, against the 10 to 30% typically reported in industrial societies.
Kenneth Wright’s team ran the complementary experiment, sending people camping for a week with no electric light. Melatonin onset shifted about two hours earlier. Sleep duration changed from 6.7 hours to 6.8 — statistically indistinguishable. Light moves when you sleep far more than how long.
The segmented sleep story is less settled than you have been told
You have probably encountered the claim that humans naturally sleep in two blocks — a “first sleep” and “second sleep” with an hour or two of wakefulness between — and that the consolidated night is an industrial invention.
Two strands support it. The historian A. Roger Ekirch documented references to first and second sleep in pre-industrial European texts, publishing in the American Historical Review in 2001. And Thomas Wehr, at the NIMH, showed in the early 1990s that volunteers placed in 14 hours of darkness developed sleep that split into two blocks separated by a waking interval of one to three hours.
Both findings are real. Neither is quite what the popular version claims.
Wehr’s condition was fourteen hours of enforced darkness, sustained for weeks — a photoperiod corresponding to deep winter at high latitude, which almost no modern human experiences. It demonstrates what the circadian system can do, not what people historically did.
And Ekirch’s thesis is under active challenge. Niall Boyce, writing in Medical History in 2023, noted that the 2001 article rested on 63 references drawn from 58 sources spanning five centuries, and that his own full-text search of early English books returned 15 hits for “second sleep,” of which only five explicitly described both a first and a second sleep. Boyce argues the phrase also meant a watch shift, or the deep early portion of a night, or simply a night that happened to be broken — and that collating ambiguous sources does not make them less ambiguous.
Ekirch replied in the same journal in 2024, reporting that he has since accumulated around 2,000 textual allusions of varying quality and at least 18 seventeenth- and eighteenth-century medical texts referring to the period after first sleep, and denying that he ever described the intervals as well-defined.
This is a live scholarly dispute between two people reading the same corpus. Neither side has a denominator — nobody knows how many pre-industrial nights went unmentioned. It is reported almost everywhere as settled fact.
Worth noting too: the Yetish measurements found no segmented sleep. Extended night waking occurred on 10 of 210 winter days and 5 of 364 summer days. Those groups live near the equator, where nights are short — exactly the condition under which Wehr’s model would predict no segmentation. The two literatures may not actually be in conflict.
How much variation is normal
A great deal, and it is stable within a person.
Hans Van Dongen’s group subjected 21 people to repeated sleep deprivation and found that between 67.5% and 92.2% of the variance in how badly they were impaired was attributable to stable individual differences — a trait, not a mood. Some people genuinely fall apart on six hours and some genuinely do not.
At the extreme, rare genetic variants produce true short sleepers. Ying-Hui Fu’s group identified a variant of the transcriptional repressor DEC2 associated with a human short-sleep phenotype in 2009, and a rare mutation affecting the β1-adrenergic receptor in 2019, with mutant mice sleeping around 55 minutes less than controls. These are genuinely rare, and almost everyone who believes they have one does not.
The J-curve, and what survives testing
Observational data consistently show a J-shaped relationship between sleep duration and mortality. Francesco Cappuccio’s meta-analysis pooled 27 cohorts, 1,382,999 participants and 112,566 deaths: short sleep carried a relative risk of 1.12, long sleep 1.30.
Note which arm is larger. Long sleep looks worse.
But when a 2024 Mendelian randomisation study used genetic variants to test causation, only the short-sleep arm survived — an estimated 0.60 years of lifespan, partly mediated through coronary disease and type 2 diabetes. The long-sleep estimate was null. That is suggestive rather than decisive, since the analysis had few genetic instruments for long sleep, but it fits the obvious explanation: illness, depression and hardship cause long sleep far more readily than long sleep causes them.
The practical reading
Chronic short sleep is genuinely bad for you, and the evidence for that is strong. Seven hours is a sensible floor.
But eight is not a requirement, waking in the night is not a malfunction, and the difference between your sleep and your neighbour’s may be constitutional rather than a failure of discipline. No trial has ever randomised people to seven, eight or nine hours for years and counted the deaths — all the duration evidence is observational or genetic.
Which means the most harmful thing about the eight-hour rule may be the anxiety it generates in people who were sleeping fine.
Sources
- Watson NF, Badr MS, Belenky G, et al. Recommended amount of sleep for a healthy adult: a joint consensus statement of the AASM and Sleep Research Society. Journal of Clinical Sleep Medicine 2015;11(6):591-592.
- Yetish G, Kaplan H, Gurven M, et al. Natural Sleep and Its Seasonal Variations in Three Pre-industrial Societies. Current Biology 2015;25(21):2862-2868.
- Wright KP, McHill AW, Birks BR, et al. Entrainment of the Human Circadian Clock to the Natural Light-Dark Cycle. Current Biology 2013;23(16):1554-1558.
- Ekirch AR. Sleep We Have Lost: Pre-industrial Slumber in the British Isles. American Historical Review 2001;106(2):343-386.
- Boyce N. Have we lost sleep? A reconsideration of segmented sleep in early modern England. Medical History 2023;67(2):91-108; and Ekirch AR, Reflections on “Have we lost sleep?” Medical History 2024;68(3):254-262.
- Wehr TA. In short photoperiods, human sleep is biphasic. Journal of Sleep Research 1992;1(2):103-107.
- Van Dongen HPA, Baynard MD, Maislin G, Dinges DF. Systematic interindividual differences in neurobehavioral impairment from sleep loss. SLEEP 2004;27(3):423-433.
- Cappuccio FP, D’Elia L, Strazzullo P, Miller MA. Sleep Duration and All-Cause Mortality: A Systematic Review and Meta-Analysis. SLEEP 2010;33(5):585-592.
- Wu Y, Zhang CY, Liu X, et al. Shared genetic architecture and causal relationship between sleep behaviors and lifespan. Translational Psychiatry 2024;14:108.
