The Science of Sleep: Nightly Routines That Work

Nearly a third of U.S. adults sleep under seven hours, and fewer than 55% wake up well-rested. Here is what large cohort studies and wearable datasets from Oura and WHOOP say actually improves sleep.

In 2024, 30.5% of U.S. adults slept six hours or less on an average night, and only 54.8% said they usually woke up feeling well-rested — figures from the CDC's National Center for Health Statistics, drawn from the 2024 National Health Interview Survey and published in 2026. That second number is the one worth sitting with: roughly half the country is spending seven or eight hours horizontal and still waking up depleted. Duration is only part of the problem.

The last decade of sleep research — especially the arrival of large wearable datasets covering hundreds of thousands of people — has shifted the emphasis. It's less about hitting a magic number of hours and more about when you sleep, how consistently, and what you put into your body and your eyes in the hours before bed. Here's what the evidence supports, with the numbers attached.

Key statistics at a glance

30.5%

US adults sleeping six hours or less a night

CDC/NCHS Data Brief 559, 2026

$411B

Annual US economic cost of insufficient sleep

RAND Europe, 2016

20–48%

Lower mortality among the most sleep-regular adults

SLEEP, 2023 (UK Biobank)

15.3 min

REM sleep lost on alcohol nights, across 600K+ Oura members

Oura, November 2025

18.1% had trouble staying asleep and 15.4% had trouble falling asleep, according to the same NCHS analysis.
Sleeping under six hours a night carries a 13% higher mortality risk than sleeping seven to nine, per that same RAND report.
400 mg of caffeine taken six hours before bed reduced objectively measured sleep by more than an hour, in a 2013 Journal of Clinical Sleep Medicine trial — and participants didn't notice.

Regularity matters more than duration

The single most useful finding for anyone building a nightly routine came out of the UK Biobank. Windred and colleagues, publishing in SLEEP in 2023, calculated a Sleep Regularity Index from over 10 million hours of wrist accelerometer data across 60,977 adults, then tracked mortality for a mean of 6.3 years.

The most regular sleepers had a hazard ratio of 0.70 for all-cause mortality versus the least regular, even after full adjustment — a 20% to 48% risk reduction across the top four quintiles depending on the model. Critically, the authors found sleep regularity was a stronger predictor of all-cause mortality than sleep duration was.

That reframes the advice. "Get eight hours" is a target most people miss and then feel bad about. "Go to bed and get up within a 30-minute window, seven days a week" is a target you can actually hit, and it appears to carry more signal.

Consumer wearable data points the same direction. WHOOP's published member analysis found that people with higher sleep-consistency scores were 4% more likely to report feeling rested on their daily surveys, while members with low consistency scores were more than twice as likely to report feeling exhausted. WHOOP also reports that members with greater consistency logged slightly more slow-wave and REM sleep, and less time awake, than members getting the same total sleep time with worse consistency.

The American Academy of Sleep Medicine and Sleep Research Society still recommend seven or more hours per night for adults 18–60, a consensus reached in 2015 by a 15-member expert panel. Treat that as the floor. Treat regularity as the lever.

What wearable data reveals about sleep across a lifetime

Oura published an analysis of more than 25,000 members in July 2025 that maps how sleep architecture changes with age — a useful reality check if you're comparing yourself to an idealized eight-hour night.

Average sleep duration fell from 7 hours 21 minutes among members aged 20–25 to 6 hours 43 minutes among those aged 80–85. Wake after sleep onset — total time awake during the night — rose from under 40 minutes in the youngest group to over 1 hour 20 minutes in the oldest. Deep sleep declined by roughly 50% across that span. Sleep timing shifted steadily earlier with age. Women slept about 16 minutes longer than men, a pattern echoed in Oura's 2025 Year in Review, which reported women averaging 7.24 hours versus 6.8 for men.

The practical read: waking once or twice at 60 is normal physiology, not a failed routine. Some of what people try to fix with sleep aids is just aging.

Light is the strongest signal you control

Your circadian system takes its primary cue from light, and evening screens are unusually good at sending the wrong one. In a controlled Harvard-affiliated study published in PNAS in 2015, Chang and colleagues had 12 adults read for four hours before bed — from a light-emitting iPad on some nights, a printed book on others, for five consecutive nights each.

Reading on the tablet suppressed evening melatonin by 55% compared with no suppression in the print condition. Participants took nearly 10 minutes longer to fall asleep, got significantly less REM sleep, and — the finding the researchers called most surprising — took longer to become fully alert the next morning.

Two things follow. First, the fix isn't only "night mode" — it's dimming the whole room, because total light intensity matters alongside wavelength. Second, four hours of pre-bed screens is an extreme dose, but it establishes the direction of the effect. A realistic target: cut bright screens 60 to 90 minutes before bed and drop household lighting to lamp level. If evening screen time is the hardest part of your routine to move, our guide to reducing screen fatigue covers the habit side of it.

Temperature, caffeine, and alcohol: the measurable inputs

Temperature. The old advice to keep your bedroom at 65°F is more confident than the evidence. Baniassadi and colleagues, publishing in Science of the Total Environment in 2023, monitored roughly 11,000 nights of sleep among 50 community-dwelling adults aged 65 and older using wearables and in-home environmental sensors. Sleep efficiency was highest between 68°F and 77°F, peaking in the 70–74°F band, and dropped 5–10% as nighttime temperature climbed from 77°F toward 86°F. Efficiency also fell below 68°F, and the authors emphasized substantial between-person variation. Treat any number as a starting point to test, not a rule.

Caffeine. Caffeine's mean plasma half-life in healthy adults is about five hours, ranging roughly 1.5 to 9.5 hours depending on genetics, smoking, liver function, and medication. In a 2013 Journal of Clinical Sleep Medicine trial, Drake and colleagues gave 12 healthy sleepers 400 mg of caffeine at zero, three, and six hours before bed. Even the six-hour dose cut objectively measured total sleep by more than an hour — and participants were largely unaware of the disruption. A cutoff of eight to ten hours before bed is a defensible default.

Alcohol. Oura's November 2025 analysis compared alcohol-tagged nights against each member's own surrounding sober nights, across more than 600,000 members:

Alcohol nights vs. each member's sober baseline (% change)

Heart rate variability	-15.6%
Resting heart rate	8.2%
Sleep Score	-6.8%
Total sleep	-5%
Deep sleep	-5%
REM sleep	-4.2%
Sleep efficiency	-2.2%

Source: Oura analysis of 600,000+ members, November 2025

Metric	Change on alcohol nights	Percent change
Total sleep	−34.6 min	−5.0%
Sleep Score	−6.55 points	−6.8%
REM sleep	−15.26 min	−4.2%
Deep sleep	−5.42 min	−5.0%
Sleep efficiency	−2.28 points	−2.2%
Heart rate variability	−10.8 ms	−15.6%
Resting heart rate	+4.35 bpm	+8.2%

The pattern matches what sleep labs have long shown: alcohol is a sedative, not a sleep aid. The 15.6% HRV drop is the number to notice — that's your autonomic nervous system working through the night instead of recovering.

A wind-down protocol worth running for two weeks

Each element traces to something above:

Fix your wake time first. Pick one, apply it seven days a week including weekends, and let bedtime settle. Regularity was the strongest predictor in the UK Biobank data, and wake time is the easier end to control.
Last caffeine 8–10 hours before bed. For a 10:30 p.m. bedtime, that's roughly 12:30 to 2:30 p.m.
Warm shower or bath, 10 minutes, 1–2 hours before bed. A 2019 Sleep Medicine Reviews meta-analysis by Haghayegh and colleagues (17 qualifying studies) found that water-based passive body heating at 104–108.5°F, timed one to two hours before bed, shortened time to fall asleep by about 10 minutes and improved sleep efficiency.
Bright screens off 60–90 minutes out; room lights down with them.
Set the bedroom cool and test it. Start in the high 60s to low 70s °F and adjust based on how you actually sleep, not on a number you read.
Skip the nightcap on nights that matter. If tomorrow requires focus, the Oura numbers above are the tradeoff you're accepting.
Get outside within an hour of waking. Morning light anchors the circadian phase everything else here depends on, and it's free.

Give it fourteen nights before judging it. Sleep habits respond on the scale of weeks, not days.

What tired actually costs

RAND Europe's 2016 analysis put the U.S. economic loss from insufficient sleep at up to $411 billion annually — 2.28% of GDP — driven by mortality risk and lost productivity, with the equivalent of 1.2 million working days lost each year, and a 13% higher mortality risk for people averaging under six hours a night versus seven to nine.

The safety numbers are starker. Williamson and Feyer's 2000 study in Occupational and Environmental Medicine, testing 39 participants across 28 hours of sleep deprivation, found that after 17 to 19 hours awake, performance on some tasks matched or fell below a 0.05% blood alcohol concentration, with response speeds up to 50% slower. The AAA Foundation for Traffic Safety's 2024 report estimated that 17.6% of all fatal U.S. crashes between 2017 and 2021 involved a drowsy driver — roughly ten times what police-reported data captures.

Sleep sits upstream of nearly every health metric an insurer looks at: blood pressure, weight, cardiovascular risk, mental health. Improving it compounds in both directions — better focus now, better numbers when you apply for coverage later. You can check your coverage options or read how term life insurance is priced against health factors like these.

Start with the wake-up time

If you take one thing from the data, make it this: the strongest single predictor in the largest study available wasn't how long people slept — it was how consistently. A fixed wake time, morning light, and a caffeine cutoff will move your sleep more than any supplement, gadget, or tracker will. The tracker is useful for telling you whether it worked.

The Science of Sleep: What the Data Says About Nightly Routines, Rest, and Focus