Researchers analysing 23 biological ageing clocks report that both short and long sleep are associated with faster ageing across the body, with the lowest ageing levels found among people sleeping roughly 6.4 to 7.8 hours a day.

What a biological ageing clock is

The phrase sounds like a metaphor and is not. An ageing clock is a statistical model that estimates a person’s biological age from measurable markers — most commonly patterns of chemical tags on DNA, but also blood proteins, immune cell profiles or organ function tests.

Each clock is built by finding markers that correlate with chronological age in a training population, then applying that model to new individuals. Someone whose markers look older than their birth certificate is described as having accelerated biological ageing.

The use of 23 different clocks is the methodologically interesting part. Individual clocks disagree with each other, and a result that holds across many of them is considerably more robust than one resting on a single measure. That is a genuine strength of this study rather than a detail.

The U-shaped curve, and why it is easy to misread

The finding is that too little and too much both look bad, with an optimum in between. That shape appears repeatedly in health research, and it carries a specific trap.

Short sleep has plausible mechanisms behind it — effects on metabolism, blood pressure, inflammation and immune function are reasonably well established. Long sleep is different. Sleeping ten hours is not obviously harmful in itself, and researchers have long suspected that in studies of this kind it is frequently a symptom rather than a cause: people sleep longer because they are unwell, depressed, in pain, or have an undiagnosed condition.

If that is what is happening, then setting an alarm to cut your sleep from nine hours to seven would not help at all. The honest position is that the association is real and the direction of causation at the long end is not established.

The precision problem

A range of 6.4 to 7.8 hours reads as though sleep should be measured to the minute. It should not.

Those figures are where a curve fitted to population data is lowest. They are not a threshold, and there is nothing that happens at 6.3 hours that does not happen at 6.5. Individual variation in sleep need is substantial and partly genetic, and self-reported sleep duration — which many such studies rely on — differs measurably from what devices record.

The useful reading is directional: consistently very short sleep is associated with worse markers, the middle of the range looks best, and nobody should be alarmed by fifteen minutes either side.

What would settle it

Observational studies of this kind can show association at scale but cannot establish cause. What would settle the question is a trial in which sleep duration is deliberately changed and ageing markers are measured before and after.

Small studies of that design exist, mostly on sleep restriction over short periods, and they do show measurable biological effects. Extending them to years is difficult, expensive and hard to justify ethically at the extremes — which is why observational data continues to carry most of the argument.

This describes research findings and is not medical advice. Persistent problems with sleep are worth raising with a doctor.

Sources

• ScienceDaily — sleep duration and biological ageing across 23 ageing clocks, 24 August 2026