A study of more than 95,000 people found that sleep duration, night-to-night variation and different stages of sleep were associated with the risk of developing a wide range of health conditions.

When we talk about sleep, one question tends to dominate: how many hours did you get?

But two people who both sleep for seven hours may experience quite different nights. One might sleep fairly continuously. The other might wake repeatedly or have much more variation in how long they sleep from one night to the next.

A study published in PLOS Medicine in September 2026 suggests these differences could matter.

Researchers analysed objectively measured sleep data from 95,559 UK Biobank participants and compared several aspects of their sleep with the health conditions they developed over the following years. [1]

This was an observational study, so it shows associations. It cannot prove that changing the way we sleep will prevent disease.

What did the researchers study?

Participants had an average age of 56 and wore a wrist accelerometer continuously for seven days.

Researchers used the data to estimate total sleep duration, REM and deep sleep, night-to-night variation in sleep duration, and time spent awake after initially falling asleep.

Participants were then followed for a median of 8.9 years, with researchers comparing these measurements with 1,049 different health outcomes. [1]

What did they find about sleep duration?

Researchers found 86 conditions where the relationship between sleep duration and risk was not simply a straight line.

For 69 of those 86 conditions, the lowest estimated risk occurred somewhere within a 6–8 hour sleep window.

Compared with people sleeping for 6–8 hours, those sleeping for less than six hours had significantly higher rates of 24 conditions after the researchers applied their strictest statistical test.

Very short sleep stood out in particular.

When researchers divided participants into five groups — less than 5 hours, 5–6 hours, 6–8 hours, 8–9 hours and 9 hours or more — they identified 41 significant associations with higher disease risk.

37 of those 41 were among people sleeping for less than five hours. [1]

These percentages describe differences between groups in this study. They do not mean that sleeping a particular number of hours will increase or decrease one person’s risk by the same amount.

What about sleeping for longer?

More sleep was not necessarily associated with lower risk.

Sleeping for more than eight hours was associated with a 60% higher relative risk of major depressive disorder compared with sleeping for 6–8 hours.

This was the only condition to remain significantly associated with longer sleep after the researchers applied their strictest statistical test, compared with 24 conditions associated with sleeping for less than six hours.

The researchers point out that longer sleep can itself occur with depression. The longer sleep could therefore be a consequence or early sign of illness rather than a cause. [1]

Does this mean we only need six hours?

No.

The NHS says adults need an average of 7–9 hours of sleep, although everyone is different. [2]

The study did not establish a new recommended amount of sleep. It found that, for many of the health conditions showing a non-linear relationship with sleep duration, the lowest estimated risk occurred somewhere between six and eight hours.

It should not be interpreted as evidence that everyone should aim for six hours.

And duration wasn’t the only aspect of sleep associated with health.

What happens while we’re asleep?

Sleep isn’t one continuous state. During a normal night we move through different stages, including lighter sleep, deep sleep and rapid eye movement (REM) sleep.

REM is particularly associated with vivid dreaming, although dreaming can occur during other sleep stages too.

The researchers used wrist movement data to estimate how long participants spent in these different stages.

These sleep-stage estimates have important limitations, but they produced some of the study’s most interesting findings.

People with relatively high amounts of estimated REM sleep had around 48 minutes more each night than people with relatively low amounts.

This difference was associated with lower rates of 83 health conditions across 12 disease categories.

The dementia figure is eye-catching, but it needs particular care.

The study does not show that gaining another 48 minutes of REM sleep would reduce someone’s dementia risk by 46%. It means that greater estimated REM sleep was associated with a lower subsequent rate of dementia after researchers adjusted for a range of other factors. [1]

What else did they find?

REM sleep wasn’t the only finding. More estimated deep sleep was associated with lower rates of seven conditions, while greater night-to-night variation in sleep duration was associated with higher rates of anxiety and depression. [1]

Sleep stages were estimated from wrist movement rather than measured in a clinical sleep study, so the results should be treated as estimates.

What can we take from this?

This was an observational study, so it cannot prove that differences in sleep caused differences in health. Poor health may also affect sleep before an illness is diagnosed.

The NHS recommends 7–9 hours of sleep for adults and, if you’re struggling to sleep, advises keeping regular waking hours, relaxing before bed, exercising regularly and avoiding tea, coffee and alcohol for at least six hours before bed. [2]

The bottom line

The study suggests that when it comes to our health, how we sleep may matter as well as how long we sleep.

But there’s no reason to start chasing REM sleep or deep-sleep scores. Give yourself enough time to sleep, and seek advice if sleep problems persist.