How much sleep do you actually need?

Everybody knows someone who claims to run fine on five hours. Maybe you are that person. The uncomfortable finding from four decades of sleep labs is that the people most confident they’ve adapted to short sleep are usually the ones performing worst — and they can’t tell, because the part of the brain that would notice is the part that’s impaired.

That’s not a scold. It’s the single most useful thing to know before you look at any chart of recommended hours, because it explains why “I feel fine” is such bad evidence. So here’s what the science actually supports, where it’s genuinely shaky, and how to work out your own number.

The official numbers, and where the experts disagree

The most-cited guideline comes from a National Sleep Foundation panel of 18 experts who reviewed 312 studies through a formal consensus process in 2015. For adults 18 to 64, they landed on 7 to 9 hours. For adults 65 and older, 7 to 8. Teenagers 14 to 17 need 8 to 10. School-age kids, 9 to 11. Newborns, a staggering 14 to 17.

Those numbers held up. In 2026 the NSF published a ten-year review covering 133 meta-analyses and reaffirmed every range without changing a single one. The review also looked for evidence that women need more sleep than men — a claim that circulates constantly online — and found none.

The chart has a second tier that gets ignored and shouldn’t. Alongside “recommended,” the panel marked ranges that may be appropriate for some people — 6 hours or 10 hours for a 30-year-old. The panel’s own view was that a handful of people live healthily outside the main range. Most people who think they’re in that group aren’t.

Meanwhile, about 30.5% of American adults slept under 7 hours a night in 2024, according to CDC data. Roughly one in three of us is running a deficit.

Is six hours enough? The study that settled it

In 2003, Hans Van Dongen and David Dinges ran the experiment everyone still cites. Forty-eight healthy adults lived in a lab for two weeks under continuous monitoring — no caffeine, no alcohol, no sneaking a nap. They were randomly assigned 8, 6, or 4 hours in bed per night for 14 straight nights. A fourth group stayed awake for three days straight.

The 8-hour group was basically flat. The 6-hour group deteriorated steadily: by the end of two weeks, their attention lapses and working memory had degraded to the level seen after one full night without sleep. The 4-hour group hit the equivalent of two nights without sleep. And the damage was still climbing at day 14 — the curves hadn’t flattened, which means nobody was adapting.

That last part is the trap. Three days of total sleep deprivation moved subjective sleepiness ratings more than 2 points. Fourteen days at 6 or 4 hours moved them about 1 point — and the rating plateaued early. People got objectively worse and worse while feeling like they’d bottomed out at “a bit tired.” Your sense of how rested you are is not a reliable instrument, which is exactly why so many people believe they thrive on six hours.

The rare people who really do need less

A small number don’t need the standard range. In 2009, Ying-Hui Fu’s lab at UCSF found a mother and daughter who slept about 6.25 hours a night, felt great, and carried a mutation in a gene called DEC2. A decade later the team found a second variant, in ADRB1, in a family who ran on roughly four hours.

These are real. They’re also vanishingly rare — best estimates put familial natural short sleepers at well under 1% of the population, and UCSF has spent years assembling a study group of only about a hundred of them. The tell isn’t willpower or coffee tolerance: true short sleepers have been this way since childhood, it runs in the family, they don’t sleep in on weekends, and they don’t nap. If you crash on Saturday morning or need an alarm to function, you’re a sleep-deprived normal sleeper, not a genetic outlier.

Why more isn’t automatically better

The relationship between sleep and mortality is a U, not a slope. Cappuccio’s 2010 meta-analysis pooled 16 studies and 1.38 million participants: short sleepers carried a 12% higher risk of death, long sleepers a 30% higher risk. But read the long-sleep side carefully — it’s the shakiest number here. Regularly sleeping ten hours is more plausibly a symptom than a cause: depression, undiagnosed apnea and early illness all lengthen sleep. The short-sleep side of the curve has experimental backing; the long-sleep side mostly doesn’t.

How to find your actual number

The clean lab version: give people an enormous sleep opportunity and see where they settle. When researchers kept subjects in bed up to 16 hours a day, young adults converged on about 9 hours and adults aged 60–72 on about 7.5 — direct evidence that sleep capacity shrinks a little with age.

You can approximate this at home over about two weeks, ideally on vacation. Go to bed when you feel genuinely sleepy. Set no alarm. Expect the first three or four nights to run long — that’s you paying off debt, not measuring need. Around day seven to ten your wake time will start landing in the same 30-minute window on its own. That settling point, held for several days, is your number.

Two honest caveats. Most people can’t take a two-week sleep vacation, so a shortened version — a long weekend plus a week of consistent early bedtimes — gets you a rough estimate, not a precise one. And if you never stabilize, or you’re sleeping nine-plus hours and still exhausted, that’s a medical question, not an arithmetic one.

Duration is only part of the story

One more finding worth holding onto. A 2024 study calculated sleep-regularity scores from over 10 million hours of accelerometer data across 60,977 UK Biobank participants and found that consistency of sleep timing predicted mortality risk better than duration did. The most regular sleepers had 20–48% lower all-cause mortality risk than the most irregular.

Seven and a half hours every night beats a chaotic average of eight. If you’re going to change one thing this week, make it your bedtime rather than your total — our sleep calculator will tell you where that bedtime should sit for the morning you need, and the sleep debt calculator shows what the current week is costing you.