Sleep and Longevity: What the Evidence Says

✓✓Strong evidence

Sleep is the only longevity intervention that costs nothing, works every night, and has no substitute. You can make up a missed workout. You cannot make up a decade of short nights.

This guide collects what human trials actually show and turns it into decisions you can make tonight. I write it from the practical side, because I have tracked my own sleep since 2019, and most of what I now do I originally dismissed as overkill.

• 7–9 hrecommended adult sleep (AASM)
• −20%insulin sensitivity after a week at 5 h
• 17–19 °Cbedroom temperature with best outcomes
• 10–14 daysfor a fixed wake time to shift your clock

How much sleep do you actually need?

Adults aged 18 to 60 need 7 or more hours per night. That is the 2015 consensus of the American Academy of Sleep Medicine and the Sleep Research Society, drawn from a review of more than 300 publications. Need is a distribution: most people sit between 7 and 9 hours, and genuine short sleepers carrying the DEC2 variant are a fraction of a percent of the population.

The mortality curve is U-shaped. Cappuccio’s 2010 meta-analysis, covering 16 prospective studies and over 1.3 million people, found rising risk both below 6 hours and above 9. The long-sleep arm is most likely reverse causation: illness lengthens sleep rather than sleep causing illness.

One distinction gets lost in most guides. Time in bed is not sleep time. If you lie down for 8 hours but take 25 minutes to fall asleep and wake three times, your actual sleep is closer to 7 hours. Sleep efficiency below 85% means your problem is quality, not duration.

What does a week of sleep restriction do to your body?

In Buxton’s 2010 trial, healthy men slept 5 hours a night for a week. Insulin sensitivity fell by about 20%, and their bodies compensated by secreting more insulin. The change happened without any change in diet or body weight.

This is not an isolated result. Back in 1999, Spiegel showed in the Lancet that six nights of four hours pushed glucose tolerance in healthy young men down to levels normally seen in the elderly. Two independent teams, two decades apart, the same direction of effect.

You do not even need a week. Donga showed in 2010 that a single night cut to 4 hours is enough to worsen insulin sensitivity in both liver and muscle. Those are two separate studies making two separate claims, and the internet routinely merges them into one.

Beyond metabolism, inflammatory markers rise, evening cortisol stays elevated, and cognitive performance drops measurably. That last part is the unsettling one: participants stop noticing their own decline after roughly three nights, while test scores keep falling. You adapt to the feeling of being tired, not to the cost of it.

What matters more, sleep regularity or sleep duration?

For decades we measured sleep with one number: hours. Recent years shifted the emphasis. In the Multi-Ethnic Study of Atherosclerosis, objectively measured sleep regularity derived from actigraphy was associated with lower mortality, independently of sleep duration.

In practice that means something that sounds trivial and almost nobody does: 7 hours every night at the same time is probably better than 8 hours on a 6-6-10-7-9 schedule. Your circadian rhythm anchors to when you wake and to morning light, not to when you go to bed.

I do have a caveat you will not find in most summaries of that study. Sleep regularity correlates with an entire lifestyle: people with steady rhythms work shifts less often, drink less in the evening, and hold more stable jobs. Cohort studies adjust for this statistically but do not eliminate it. My bet is that the effect is real but smaller than the headlines. I cannot prove that.

What improves sleep most for the least effort?

A fixed wake time

Pick one time and hold it seven days a week, with roughly twenty minutes of tolerance. Shifting your wake time two hours at the weekend does to your body roughly what flying across two time zones does, which is why it is called social jet lag.

The effect shows up after 10 to 14 days, because that is how long the suprachiasmatic nucleus takes to reset. This is the only one of the three that shifts your circadian rhythm on its own.

A bedroom at 17 to 19 °C

Falling asleep requires core body temperature to drop by roughly 1 °C. Your body does this by dilating peripheral blood vessels, dumping heat through hands and feet. In an overheated room there is nowhere for that heat to go.

Okamoto-Mizuno’s 2012 review shows that high ambient temperature increases awakenings and shortens slow-wave sleep, with high humidity amplifying both. This is the most common and the cheapest problem to fix among people who ask me about sleep.

Daylight within the first hour of waking

In Wright’s 2013 study, a week of camping without artificial light shifted melatonin onset about two hours earlier and synchronised participants to sunrise. Intensity is what matters: an overcast day outdoors gives 1,000 to 5,000 lux, a well-lit living room gives 100 to 300.

Ten minutes on a balcony beats an hour by an indoor window. Glass strips out most of the effect by cutting both intensity and part of the spectrum.

How do you measure whether your sleep actually improved?

Without measurement you cannot separate an intervention from a good week. You do not need a £300 ring for this. Two numbers written down each morning are enough:

1. How many minutes it took you to fall asleep, estimated to the nearest five.
2. How many times you woke up.

Trackers like Oura and Whoop are decent at detecting whether you are asleep and noticeably weaker at classifying sleep stages. The agreement figures manufacturers quote usually refer to sleep-wake detection, not stage classification. Treat the hypnogram as orientation, not diagnosis.

A practical rule: read weekly trends, not single nights. One bad night according to an app means nothing. Four in a row means something.

When does a sleep problem need a doctor rather than hygiene?

If after two weeks of consistent basics you still take more than 30 minutes to fall asleep, or wake unrefreshed after 8 hours in bed, stop looking for another technique. That is the point to screen for sleep apnoea, restless legs, thyroid dysfunction or iron deficiency.

Snoring with breathing pauses, morning headaches and daytime sleepiness despite long sleep is a combination that belongs in a doctor’s office, not a supplement shop.

What we do not know

We do not know how much of the benefit comes from sleep duration versus regularity, because regularity research is younger and largely observational. We do not know whether improving sleep architecture in someone already sleeping 7 to 8 hours moves hard endpoints. We also do not know how much an individual’s need shifts with age, because almost all the data is cross-sectional rather than longitudinal.

Where the evidence is missing, we say so. The evidence scale used across this site is described in the supplements guide.

Where to start tonight

Set one wake time for the next two weeks and air out your bedroom before bed. Those two changes cost nothing and carry the strongest evidence on the list. Everything else gets added in sequence, following the sleep protocol, and the parts of conventional sleep hygiene that turn out to be myth are taken apart in sleep hygiene.