Is 8 hours of sleep a myth? What the mortality data actually shows

Eight hours is a round number, not a biological law. When researchers pool the big population studies that track how long people sleep and how long they live, the lowest-risk amount lands closer to seven hours than eight, and the relationship is U-shaped: both short and long sleep are associated with a higher risk of dying [s1][s2]. So "everyone needs eight" overstates a target that the mortality data put a little lower — and, just as importantly, misreads what that data can tell you.

The more useful way to hold this is that the numbers describe a population, not a person, and describe an association, not an instruction.

What the pooled mortality data says

The foundational analysis, published in the journal Sleep in 2010, brought together 16 studies yielding 27 independent cohort samples — 1,382,999 participants followed for between 4 and 25 years, with 112,566 deaths recorded [s1]. Short sleep was associated with a greater risk of death (relative risk 1.12, 95% CI 1.06 to 1.18, P<0.01), and long sleep with a larger one still (relative risk 1.30, 95% CI 1.22 to 1.38, P<0.0001) [s1]. Notably, the long-sleep signal was the stronger of the two — the opposite of the "more is better" intuition behind chasing eight or nine hours [s1].

A 2017 dose-response meta-analysis in the Journal of the American Heart Association sharpened the shape of the curve [s2]. Across the prospective cohorts it pooled, risk was lowest at about seven hours of sleep a day, and rose on either side [s2]. For all-cause mortality, each hour below seven was associated with a 6% higher risk (relative risk 1.06, 95% CI 1.04 to 1.07), while each hour above seven carried a 13% higher risk (relative risk 1.13, 95% CI 1.11 to 1.15) [s2]. The same U-shape held for cardiovascular disease — 1.06 per hour below and 1.12 per hour above — and for stroke, where the penalty for long sleep was steepest, at 1.18 per additional hour (95% CI 1.14 to 1.21) [s2].

Why "get more sleep" is the wrong lesson from the long-sleep half

Here is the trap. It is tempting to read the U-curve as proof that sleeping nine or ten hours harms you, and therefore that you should force yourself to sleep less. The cohorts cannot support that. They are observational: they record how long people sleep and what happens to them, but not why they sleep that long [s1][s2]. Long habitual sleep is a well-known marker of underlying illness — depression, chronic disease, inflammation, undiagnosed conditions that both lengthen sleep and shorten life — so the raised risk on the long side may reflect who sleeps eleven hours rather than the hours themselves [s2]. Both meta-analyses flagged significant heterogeneity between the studies they combined, another reason to treat the precise numbers as directional rather than exact [s1][s2].

The short-sleep side is on firmer causal ground, because there are plausible mechanisms and experimental support for the harms of genuine sleep restriction. But even there, the association in these studies is modest, and it does not follow that nudging a comfortable seven-hour sleeper up to eight buys them anything.

So how much should you actually aim for?

The practical reading is undramatic. Somewhere in the range of about seven hours is where the population risk is lowest, and there is no evidence that manufacturing an eighth or ninth hour extends life [s1][s2]. Individual need varies, and the right amount is the one that leaves you rested and functioning, not a figure copied from a slogan. If you routinely sleep far less than seven, that is worth attention; if you consistently need far more than eight to feel normal, the sleep length is less likely to be the problem than a reason to ask what is driving it.

What is settled, and what is not

What the data settle is narrow but real: across more than a million people, seven hours or so sits at the bottom of the risk curve, and both extremes are associated with higher mortality, with long sleep carrying the larger association [s1][s2]. What they do not settle is causation in either direction, how much of the long-sleep risk is really a sickness signal, or what any single person's optimum is — none of which a slogan about eight hours was ever equipped to answer.

For where the official guideline number comes from, see the sleep guideline that never said eight. This article describes evidence and is not medical advice.

Sources

• Sleep Duration and All-Cause Mortality: A Systematic Review and Meta-Analysis of Prospective Studies — Sleep, 2010-05-01
• Relationship of Sleep Duration With All-Cause Mortality and Cardiovascular Events: A Systematic Review and Dose-Response Meta-Analysis of Prospective Cohort Studies — Journal of the American Heart Association, 2017-09-09