Sleep and longevity

Yes, both too little and too much sleep are linked to dying earlier, and the risk is lowest at about 7 hours a night. A meta-analysis pooled about 1.4 million people and compared them with those sleeping about 7 hours. Short sleep (usually under 6 to 7 hours, depending on the study) went with a 12 percent higher risk of dying early.^[1] For long sleep (usually over 8 to 9 hours), the risk was 30 percent higher.^[1] Drawn as a graph, those numbers make a U-shaped curve. Its safe middle fits the guideline range of 7 to 9 hours for adults and 7 to 8 hours from 65.^[55]

The long-sleep end of that curve needs a second look, because illness can cause long sleep rather than the other way round. Depression or a health problem nobody has diagnosed yet can keep people in bed longer. The authors of the meta-analysis name this reverse causation as a likely part of the link.^[1] So the curve shows where the risk sits, not that an extra hour in bed harms a healthy sleeper.

The short end is easier to believe once you look at what the body does at night, and the brain is where researchers have looked hardest. In mice, the spaces between brain cells widened during sleep, and tracer dyes washed out about twice as fast as in waking mice.^[2] In people, brain scans show slow waves of spinal fluid moving through the brain during deep sleep.^[3]

That cleaning picture is not settled, because a 2024 mouse study using a different method found less clearance during sleep, not more.^[4] The fluid clearly moves differently at night. How much waste it removes, and whether that includes the proteins tied to Alzheimer's in humans, is still argued over. Headlines turned this into a brain that cleans ten times harder at night, which is a bigger number than any of these studies reported.

Memory stands on firmer ground than brain cleaning. Deep sleep (also called slow-wave sleep, when the brain produces big, slow waves) helps the brain store what you learned during the day. When researchers strengthened those slow waves with soft, precisely timed sounds, people remembered word pairs they had learned before sleep better the next morning.^[84] Small studies in older adults point the same way,^[85] although as of October 2026 we found no FDA-cleared, peer-reviewed consumer device that does this. In older adults, the natural loss of deep sleep also goes along with worse overnight memory.^[9]

Growth hormone is tied to the same stage. In men, most of it is released in a pulse during the first deep-sleep phase, shortly after falling asleep, and nightly output falls with deep sleep as men age.^[8] Repair is where the evidence turns to animals. In zebrafish neurons, sleep increased chromosome movement and reduced the DNA damage that builds up while awake, though two other cell types showed no such effect.^[70] For humans that is a hint, not proof. Most of these night jobs are tied to particular sleep stages, and those stages are spread very unevenly across the night.

A night of sleep begins when sleep pressure outweighs your body clock, and from then on your brain cycles through four stages in rounds of roughly 90 minutes. Sleep researchers have described this tug of war between two forces since the early 1980s.^[5] Sleep pressure builds the longer you are awake, like a balloon slowly filling through the day. Your body clock pulls the other way, with an alerting signal that runs through the day and eases off at night.

Sleep pressure has a molecule behind it. Adenosine, a tiredness signal, builds up in the brain during long waking hours and drains away during recovery sleep.^[6] Caffeine works by blocking adenosine from docking, which muffles the tiredness signal rather than adding energy.

Once you are asleep, the four stages take turns: N1, N2, N3 and REM. N1 is the light drift from awake into asleep, when a small noise can still bring you back. N2 is the workhorse stage and fills the largest share of the night in healthy adults.^[7] N3 is deep sleep, when the brain rolls out big, slow waves and you are hardest to wake. REM (rapid eye movement) is the dreaming stage, when your brain looks almost as busy as when you are awake. Your body goes briefly limp in REM, so you do not act out your dreams.

Those stages are not spread evenly across the cycles. Deep sleep is front-loaded, so it dominates the first couple of cycles and often fades out by the last one. REM is back-loaded and grows longer as the night goes on, which puts your longest dream stretch just before you wake. A night cut short at the end therefore costs you mostly REM. The popular trick of timing your alarm to the end of a 90-minute cycle is shakier than it sounds. Cycle length varies between people and even within one night, so the alarm is aiming at a target that keeps moving.

That uneven pattern shifts further with age. A large meta-analysis found that, as adults get older, total sleep drops by about 10 minutes per decade and falling asleep takes longer.^[7] The light stages take up more of the night, deep sleep and REM shrink, and waking in the night becomes more common.^[7] Deep sleep drops fastest early on. In healthy men, it fell from about 19 percent of sleep at ages 16 to 25 to about 3.4 percent at ages 36 to 50, then levelled off.^[8]

So a long, unbroken night with plenty of deep sleep is typical of young adults, and by 60 a more broken night is normal.^[7] Researchers argue that the brain's ability to produce deep sleep wears down with age while the need for sleep does not.^[9] Protecting the deep sleep you still get, with the levers in our deep sleep guide, is therefore the practical goal as the years add up.

Poor sleep is linked to faster aging on several fronts, though most of the human data are observational and the effects are modest. The link runs in both directions, because aging itself brings lighter, more broken nights.^[7] That makes it hard to say how much of the connection starts with the sleep. The firmest evidence comes from experiments that cut people's sleep on purpose and then measured their blood sugar.

Blood sugar reacts within days

In those experiments, short sleep quickly made the body worse at handling sugar. One week of 5-hour nights lowered insulin sensitivity in healthy young men by 11 to 20 percent.^[10] Insulin sensitivity describes how well the body responds to insulin, so their bodies had to work harder to keep blood sugar in check. An older study went further, with six nights of 4 hours in a small group of young men. Sugar left their blood about 40 percent more slowly than after they had caught up on sleep, a pattern the authors compared with normal ageing.^[11] Both studies cut sleep for only a few nights in small groups of young men. They show how fast the system reacts, not what years of short sleep do.

Inflammation rises after several short nights

Inflammation follows a similar pattern, and here the number of nights matters. In lab studies, several nights of about 4 hours of sleep raised blood markers of inflammation, while a single bad night did not.^[69] Those studies measured blood markers over days rather than disease over years. They show that repeated short sleep stirs up inflammation quickly, not that it causes illness later.

Short sleep in midlife is linked to dementia

For the brain, the evidence comes from following people for decades. The Whitehall II study tracked about 8,000 British civil servants for 25 years. In that group, sleeping 6 hours or less at age 50 was linked to a 22 percent higher dementia risk later on.^[12] The study was observational, and early brain changes may also shorten sleep years before a diagnosis, so cause and effect are not settled.

Telomeres and clock genes show smaller links

The cellular markers of aging show smaller and less consistent links. Telomeres are the protective caps on the ends of your chromosomes, and a large 2025 meta-analysis found only small links between poor sleep quality and shorter telomeres.^[67] Those links held up for only a few measures, such as an overall sleep-quality score and time spent awake after first falling asleep.^[67] Many of the studies were also inconsistent or too small.^[67]

Epigenetic tags, the chemical markers on DNA that change with age, tell a similar story. In a small study of healthy men, one night without sleep changed these tags on body-clock genes in fat tissue.^[68] It also lowered the activity of clock genes in muscle.^[68] Nobody has shown that this moves an epigenetic age test, so a single bad night is a long way from the number on a biological-age test.

The clearest harm in all of this, on blood sugar and inflammation, came from cutting sleep short for days. Sleeping in at the weekend does not simply undo it. In a lab study, people who slept 5 hours on weekdays and slept in at the weekend still lost insulin sensitivity.^[74] Enough sleep night after night, one of the levers that slow aging, is therefore a more reasonable plan than counting on weekend catch-up.

According to the National Sleep Foundation, adults aged 18 to 64 need 7 to 9 hours of sleep a night, and adults from 65 need 7 to 8 hours.^[55] The American sleep-medicine societies set the floor for adults at 7 hours,^[56] and the mortality curve is lowest around 7 hours.^[1] The lower end of the range is therefore where most of the evidence meets.

Source: National Sleep Foundation.^[55]

About 7 hours also looked best for the brain. UK Biobank, a large British health study, followed about half a million adults aged 38 to 73. Those sleeping about 7 hours had the sharpest thinking and the best mental health.^[50] Both shorter and longer sleepers did worse. Like the mortality curve, this comes from watching people rather than from experiments. It shows where people did best without proving that adding or cutting an hour would change your own score.

Hours are only part of the measure, because the timing of sleep counts too. Another UK Biobank study tracked sleep timing with wrist sensors in about 61,000 adults. The more regular sleepers had about a 20 to 30 percent lower risk of dying during follow-up than the least regular ones.^[48] Regularity also predicted that risk better than hours slept did.^[48] The study was observational, so it shows a link rather than proof, but the simplest version of the fix costs nothing: a fixed wake time, weekends included.

Quality is the third measure, and the simplest number for it is sleep efficiency: the time you actually slept divided by the time you spent in bed. Seven hours asleep over eight hours in bed gives 87.5 percent. Sleep clinics often use about 85 percent as a working mark when they adjust schedules in CBT-I, the structured insomnia therapy.^[30] Treat it as a reference point rather than a proven health target. Most healthy sleepers fall asleep within about 30 minutes, so dropping off the moment your head touches the pillow is more often a sign of sleep debt than a talent.

Signs your sleep is good enough:

• You usually wake up rested, sometimes before the alarm.
• You keep your energy through the day without leaning on caffeine.
• You usually fall asleep within about 30 minutes.
• You rarely lie awake in the night.

Signs it isn't:

• You need an alarm and still struggle to get up.
• You rely on caffeine to function or crash in the afternoon.
• You fall asleep the instant your head hits the pillow.
• You feel drowsy in meetings or at the wheel.

Drowsiness at the wheel is the one sign to act on right away. If you feel drowsy while driving, stop driving and get checked soon. Sleep apnea, where breathing keeps stopping during sleep, is the most common medical cause of daytime sleepiness. It raises the risk of drowsy driving two to three times.^[81]

For everything else on those lists, a two-week sleep diary shows which measure is your problem before you change anything. For 14 days, note your bedtime, wake time, last caffeine, alcohol and how rested you feel on a scale of 1 to 5. A notebook works, and so does the sleep log of a wearable. At the end, work out your average sleep efficiency and how long you usually take to fall asleep.

Those two numbers tell you where to start. If falling asleep regularly takes more than 30 minutes on at least 3 nights a week, talk to a doctor. That number is worth a conversation, though not a diagnosis on its own. If the diary shows your bedtime sliding later on free days, the body clock is the place to look next.

Your sleep timing is set mostly by light and temperature, which is why you cannot simply decide to become a morning person. A small cluster of cells deep in the brain, the SCN, runs your 24-hour rhythm. It reads the time of day from special light-sensitive cells in your eyes.

Light can move that clock by hours, but not equally in both directions. In one experiment, a 6.7-hour block of bright light could pull the clock up to about 2 hours earlier.^[14] Given at another time, it pushed the clock up to about 3.6 hours later.^[14] Your clock slides later far more easily than it moves earlier. That fits the everyday experience that staying up late is easy and getting up early is not, as most teenagers could have told the researchers.

Morning daylight is the strongest of those light signals, because outdoor light is far brighter than normal indoor light. When researchers sent volunteers camping for a week, their light exposure rose more than fourfold, and their body clocks lined up with sunrise and sunset.^[62] Light levels are measured in lux, and for the body clock what counts is the light reaching your eyes, weighted the way the clock-setting cells see it. An expert consensus recommends at least 250 lux of that light at eye level during the day.^[13] In practice, that means getting outside soon after waking, ideally without sunglasses, since a seat by the window gives you less light than stepping outside.

Indoor light is weak compared with daylight, but it still affects the clock, especially in the evening. Light at night suppresses melatonin, the hormone that tells your body it is nighttime, and the brighter the light, the stronger the effect. Ordinary room light of around 100 lux already halves it.^[15] In another study, people read on a glowing eReader for about 4 hours on five evenings in a row. Compared with reading a printed book, this delayed their melatonin by about 1.5 hours and left them groggier the next morning.^[16]

The same consensus therefore suggests keeping clock-setting light at 10 lux or less in the 3 hours before bed, far below normal room light.^[13] While you sleep, the target is 1 lux or less.^[13] Blue-blocking glasses may help a little, but dimming the whole room works better.

Temperature is the clock's second lever. In the evening your body sheds heat through your hands and feet, where blood vessels near the skin open up. How warm your hands and feet were compared with your trunk predicted how fast people fell asleep.^[17] It did so better than core temperature, melatonin or how sleepy they felt.^[17] Warm hands and feet are therefore a good sign that sleep is near. In a lab study, a thermosuit warmed the skin of older adults by about 0.4 °C without changing their core temperature. That small change almost doubled their deep sleep and cut early-morning waking.^[28]

Warm skin does not mean a warm room, though, and the bedroom itself can be cooler than many people keep it. A review found that under normal bedding, room temperatures from about 13 to 23 °C barely changed sleep stages.^[18] Heat did, especially humid heat, which cut both deep sleep and REM.^[18] Lean cool, then: around 16 to 20 °C (61 to 68 °F) is a sensible rule of thumb rather than a lab-tested number. Socks help if cold feet keep you awake.

Cooling mattresses try to do the same job by drawing heat away from the body. In a trial across three sleep centres, a heat-absorbing mattress added about 7.5 minutes of deep sleep.^[59] One author worked for the mattress maker, and claims from other makers have little independent testing, so a cool room is the free first step.

Your chronotype, whether you are naturally an early bird or a night owl, sets limits on all of this. It is partly genetic, so light and darkness can nudge it but not fully override it. Fighting it shows up in the data. Social jetlag, the gap between your sleep timing on free days and on work days, was linked to a higher BMI beyond what sleep duration explained.^[52] Where your work allows it, building your day around your chronotype is a reasonable choice rather than a lack of discipline.

Caffeine, alcohol, nicotine, cannabis and late heavy meals all make sleep worse, and a simple working rule covers most of them: stop at least 3 hours before bed. Caffeine needs more room than that, because people clear it at very different speeds.

Its half-life (the time your body needs to clear half of it) ranges from about 2 to 10 hours between people.^[19] In a study measured at home with a portable sleep monitor, 400 mg of caffeine taken 6 hours before bed still cut total sleep by more than an hour.^[19]

Since 6 hours was not enough there, a cutoff about 8 hours before bed is a safer start, and about 10 hours if you are sensitive. With an 11 PM bedtime, that means no caffeine after about 3 PM, or after about 1 PM if you are sensitive. Decaf is not zero either, since a lab analysis found 3.0 to 15.8 mg of caffeine in a single shot of decaf espresso.^[61]

Alcohol works the other way round, helping you fall asleep and then breaking up the night. A review of the research found that alcohol speeds up sleep onset and increases deep sleep in the first half of the night.^[20] In the second half, as it clears your blood, sleep fragments, and REM, held down early, comes back later.^[20]

That broken second half often shows up the next morning. Wearables tend to show a higher overnight resting heart rate and a lower HRV (heart-rate variability, a rough read on overnight recovery) after drinking. Alcohol also relaxes the muscles of the airway and can make sleep apnea worse. A sensible rule is to finish at least 3 hours before bed and keep it to one drink on nights when sleep matters. Skip it if you have apnea.

Nicotine is a stimulant, even when a cigarette feels relaxing. A review found that it makes you take longer to fall asleep, breaks up the night, cuts total and deep sleep and suppresses REM.^[21] Quitting brings a stretch of worse sleep before things improve.^[21] If you still use nicotine, keep the last dose at least 3 hours before bed, and expect some rough nights when you stop.

The evidence on late meals comes mostly from people who already have sleep apnea. Among them, habitual late eaters had slightly more breathing pauses per hour, took longer to fall asleep, lay awake longer in the night and got less REM.^[22] The link is observational and the effects were small, but they all pointed the same way. Finishing your last big meal about 3 hours before bed is therefore a reasonable habit.

Cannabis trades a faster start for a worse night later on. A review found that THC, the part of cannabis that makes you high, can help you fall asleep faster but suppresses REM.^[23] Tolerance builds, so the benefit fades, and stopping brings insomnia and vivid dreams as REM bounces back.^[23] A small placebo-controlled trial in people with insomnia showed the REM cost directly. There, a single oral dose of 10 mg THC plus 200 mg CBD cut REM sleep by about 34 minutes.^[60] It also shortened total sleep by about 25 minutes.^[60]

CBD on its own looks more benign in early research, though that evidence is thin.^[23] If you use THC, keep the dose low, do not use it every night, and do not drive afterwards.

Sleeping pills are a treatment for insomnia rather than an evening habit, so the CBT-I section covers them, together with how to come off them safely. Everyday medicines are a different matter, because several common ones can cause insomnia as a side effect.

Beta-blockers, for example, lower the body's nighttime melatonin. In a small trial in people taking them, 2.5 mg of melatonin nightly for 3 weeks added about 36 minutes of total sleep.^[63] SSRIs, a common class of antidepressants, delay and reduce REM sleep, while steroids can cause insomnia, especially when taken in the evening. ADHD stimulants can delay sleep too, and the safe timing depends on the preparation, so ask your prescriber. Whatever the drug, if insomnia starts within weeks of a new medication, suspect the medication, and move or change the dose only after talking to your doctor.

Exercise and a warm bath before bed both have trial evidence for slightly better sleep, while sauna and cold plunges have little direct sleep evidence. The last two also come with safety rules of their own.

Exercise helps sleep a little, and evening sessions are mostly fine. A meta-analysis of single evening workouts in healthy adults found that exercise slightly increased deep sleep and reduced N1, the lightest stage.^[26] The exception was hard exercise ending within about an hour of bed, which may delay sleep and lower sleep efficiency.^[26]

A large observational study of wearable users points the same way.^[58] The later and harder people exercised in the evening, the later they fell asleep and the shorter and worse their sleep was. Sessions ending at least 4 hours before bed showed no such link.^[58] The wearable maker co-authored that study, but together the two suggest keeping very hard sessions away from bedtime rather than avoiding evening exercise altogether. Our exercise guide covers how to build the habit.

Morning exercise can also move the body clock. In inactive young adults, 5 days of morning exercise shifted the clock about 0.6 hours earlier in total, while evening exercise did little on average.^[57] Late chronotypes shifted earlier with both.^[57] For a night owl who wants an earlier bedtime, morning training is a reasonable extra lever.

A warm bath or shower helps too, and it is the cheapest option here. In a review of the studies, a warm bath or shower at about 40 to 42.5 °C (104 to 108.5 °F) shortened the time to fall asleep and improved sleep quality.^[27] It worked when taken 1 to 2 hours before bed, for as little as 10 minutes.^[27] The authors still called the evidence limited.

The mechanism runs backwards from what you might expect. Warm water opens the blood vessels near your skin, so when you step out into a cooler room, heat leaves your body quickly. Your core temperature then falls, which is the signal your body uses to start sleep.^[27] The recipe follows from that: about 10 minutes in water at 40 to 42.5 °C, finished 1 to 2 hours before lights-out. After that comes a cool, dim bedroom like the one in the body-clock section.

Sauna has much better evidence for the heart than for sleep. In a Finnish study of middle-aged men, 4 to 7 sauna sessions a week were linked to about 50 percent lower risk of dying from heart disease than one session a week.^[51] That study was observational. The direct sleep evidence is thin and mostly self-reported. Still, the heat-then-cool pattern resembles the warm bath, so finishing a session an hour or two before bed is a reasonable guess. Skip the sauna after drinking. If you have unstable heart disease or low blood pressure, check with your doctor first, because heat can trigger heart strain, fainting and falls.^[79]

Cold plunges are the one habit here that may work against sleep in the evening. In a lab study, an hour in 14 °C water raised norepinephrine, an alertness chemical, about sixfold.^[49] That study did not measure sleep, so the idea that a late plunge keeps you awake is a reasonable guess rather than a tested rule. If you plunge at all, morning or midday is the safer slot. Cold water can also trigger a cold-shock response that strains the heart and can cause drowning.^[80] If you have heart disease, talk to your doctor first, and never plunge alone.^[80]

For long-term insomnia, the first-line treatment is a structured program called CBT-I (cognitive behavioral therapy for insomnia), not a sleeping pill. The American Academy of Sleep Medicine (AASM) recommends it,^[30] and so do the European Sleep Research Society, the UK's NICE and the American College of Physicians. Pills come second and are usually meant for limited periods. Their benefit tends to fade once you stop, while the gains from CBT-I last.^{[29, 54]}

CBT-I works on the numbers people care about. Across 20 trials, it cut the time to fall asleep by about 19 minutes and the time awake in the night by about 26 minutes.^[29] It also raised sleep efficiency by about 10 percentage points, and the gains held at follow-up.^[29] Those gains outlast pills, too. In a trial that followed patients for two years, 44 to 63 percent were in remission, meaning their insomnia had cleared by the trial's measures.^[54] People who stopped the sleeping pill zolpidem while doing CBT-I did better than those who kept taking it as needed.^[54]

CBT-I is a short course of sessions built from a few parts. Stimulus control retrains your brain to link the bed with sleep instead of with lying awake. Sleep restriction cuts your time in bed on purpose so that sleep packs together. Cognitive work catches the catastrophic thoughts about sleep ("I'll be useless tomorrow"), relaxation training calms the body, and sleep hygiene covers basics such as caffeine, alcohol and light. The AASM guideline backs stimulus control, sleep restriction and relaxation as single treatments too, but it recommends against sleep hygiene on its own.^[30]

Stimulus control

Stimulus control comes down to a few rules you keep every night:

• Go to bed only when you are sleepy, not just tired.
• Use the bed for sleep and sex only.
• If you are still awake after about 20 minutes, get up, do something quiet in dim light, and go back when you feel sleepy.
• Get up at the same time every day, weekends included.
• Skip daytime naps while you are doing the program.

Sleep restriction

Sleep restriction feels backwards, because you start by spending less time in bed. The details vary between clinics, and the numbers below are one common example rather than a fixed rule from the guideline:

1. Keep a sleep diary for 7 to 14 days and work out your average actual sleep time.
2. Set your time in bed to that average plus 30 minutes, but never below 5 hours. Fix your wake time first, then count back to your bedtime.
3. Hold that window for a week. You will feel sleepy, which is the point, because it builds sleep pressure.
4. Adjust the window weekly by sleep efficiency, the principle the AASM guideline describes.^[30] In this example: at 85 percent or more, go to bed 15 to 30 minutes earlier; at 80 to 84 percent, hold steady; below 80 percent, cut 15 minutes.
5. Keep going until you sleep enough and are satisfied with your sleep.^[30]

Sleep restriction makes you sleepier at first, so don't drive or operate machinery when drowsy. Do it with a clinician, not alone, if you have bipolar disorder, epilepsy, a job where sleepiness is dangerous, or strong daytime sleepiness.^[30]

Even a short version of the program can work. In a four-session program for older adults with chronic insomnia, 55 percent were in remission after treatment.^[31] An app can deliver the program as well, which helps when a therapist is hard to find. In a large Norwegian trial, 58 percent of people using digital CBT-I had a clear drop in their insomnia score, a large effect compared with patient education.^[32] In Germany, the DiGA directory (diga.bfarm.de) lists insomnia apps that doctors can prescribe on statutory insurance (as of October 2026).

If a pill is needed

Sleeping pills have a place when CBT-I is not enough or not available, but the evidence for them is weaker. The AASM's 2017 drug guideline gave only weak recommendations for every sleeping-pill class.^[24]

The older drugs carry known risks. Benzodiazepines and the Z-drugs (zolpidem, zaleplon, eszopiclone) cause tolerance, falls and half-asleep behaviours. In April 2019 the FDA put its strongest warning, a boxed warning, on the three Z-drugs. The trigger was reports of serious injuries and deaths from sleepwalking, sleep-driving and other things people did while not fully awake. In an observational analysis, people prescribed sleeping pills died at three to five times the rate of non-users, and the rate rose with dose.^[53] That link does not prove the pills caused the deaths.

The newer orexin blockers (DORAs) such as daridorexant work differently, by blocking orexin, the brain's stay-awake signal. In two 3-month trials, daridorexant 50 mg reduced time awake at night by about 18 to 23 minutes more than placebo.^[25] It also shortened the time to fall asleep by about 11 to 12 minutes more.^[25] The 25 mg dose had smaller effects, and only 50 mg clearly improved daytime sleepiness.^[25] A 40-week extension found no signs of rebound insomnia or withdrawal,^[72] though the manufacturer funded the trials.

Daridorexant still comes with rules. In the EU, daridorexant (Quviviq) is prescription-only (as of October 2026).^[78] It is not for people with narcolepsy or for those taking strong CYP3A4 inhibitors, certain other medicines your pharmacist can check. Treatment should be kept as short as possible and reviewed within 3 months, and because it can cause sleepiness, be careful driving.^[78] In the UK, NICE guidance TA922 recommends it only when CBT-I has been tried and has not worked, or is not available or suitable (as of October 2026).

Which pill to take, and for how long, is a decision for your doctor, and coming off one needs the same care. Never stop benzodiazepines or Z-drugs abruptly after regular use; plan the reduction with your doctor. For benzodiazepines, withdrawal can be severe and, rarely, life-threatening (seizures, confusion).^[73]

A 2025 joint guideline on benzodiazepines suggests starting with reductions of about 5 to 10 percent every 2 to 4 weeks.^[73] Cuts are usually not more than 25 percent every 2 weeks.^[73] The pace is adjusted to how you feel, and for many people the taper takes months.^[73] That guideline does not cover Z-drugs, so for zolpidem and similar pills your doctor sets the pace individually. Run CBT-I alongside the taper.^[54] A switch to another sleep drug does not replace a supervised taper.

When to bring in a specialist

See a sleep specialist if 6 to 8 weeks of structured CBT-I have not helped or your days are badly impaired. The same goes if you also have signs of sleep apnea, restless legs or REM sleep behavior disorder, or if you have depression or work shifts. A specialist also helps if you need support tapering off benzodiazepines or Z-drugs. If you have thoughts of suicide, get help now: call 112 in an emergency, or in Germany the TelefonSeelsorge on 0800 111 0 111 or 0800 111 0 222 (free, 24/7).

See a doctor soon if you snore loudly with pauses in breathing, act out your dreams, nod off at the wheel, or have had insomnia for 3 months or more. Some sleep problems do not respond to better habits or CBT-I, and these red flags show which ones and what each calls for:

• Loud snoring, pauses in breathing that someone has seen, and daytime sleepiness → tests for sleep apnea.
• Acting out your dreams → see a neurologist (REM sleep behavior disorder, below).
• Nodding off behind the wheel → stop driving until you've been assessed, and get urgent tests for apnea or narcolepsy.^[81]
• Insomnia lasting 3 months or more → CBT-I, not a nightly sleeping pill.
• Sleepy all day despite enough time in bed, with an Epworth score of 10 or more (a short questionnaire on how easily you doze off in everyday situations) → sleep clinic.
• Unpleasant evening urges to move your legs → have your iron stores checked and talk to your doctor about treatment.^[37]
• Snoring with morning headaches and blood pressure that won't come down with treatment → strongly suggests apnea.
• Thoughts of suicide → get help now: call 112 in an emergency, or in Germany the TelefonSeelsorge on 0800 111 0 111 or 0800 111 0 222 (free, 24/7).

Insomnia counts as a disorder when trouble sleeping happens on at least 3 nights a week for at least 3 months. It also has to affect your days and persist despite enough time to sleep. Its treatment is in the CBT-I section, so the conditions below are the ones that need more than CBT-I.

Sleep apnea

Obstructive sleep apnea, in which the airway keeps collapsing during sleep, is more common than most people think. Moderate-or-worse apnea affects about 10 percent of men aged 30 to 49 and about 17 percent of men aged 50 to 70.^[33] In women of the same ages, the figures are about 3 and 9 percent.^[33] Extra weight, a thick neck, being male, older age and a recessed lower jaw raise the odds.

A quick screen is the STOP-BANG checklist. It asks about loud snoring, daytime tiredness, observed pauses in breathing and high blood pressure, then about a BMI over 35, age over 50, a neck over 40 cm and male sex. Three or more yes answers mean moderate risk, and five or more mean high risk. A diagnosis needs an overnight sleep study in a lab (polysomnography) or, for straightforward cases, a take-home test.

CPAP, a mask that keeps the airway open with gentle air pressure, is the standard treatment. If you cannot tolerate it, or your apnea is mild to moderate, a mouthguard that moves the jaw forward is an option. Sleeping position, weight loss and an implant that stimulates the tongue nerve are others. Two large trials, SAVE,^[35] and ISAACC^[46] found that CPAP did not cut heart attacks and strokes. In both, though, people wore the mask only about 3 hours a night on average, and the patients were not sleepy to begin with. CPAP therefore stays the first-line treatment when apnea causes symptoms.

Weight-loss medication has become an option too. In a 52-week trial in people with obesity and apnea, tirzepatide cut breathing pauses and shallow breaths.^[34] The drop was 20 to 24 per hour of sleep more than with placebo.^[34] On 20 December 2024 the FDA approved it (as Zepbound) for moderate-to-severe sleep apnea in adults with obesity. It is the first drug approved for the condition itself (as of October 2026). In the EU, tirzepatide (Mounjaro) is authorised for type 2 diabetes and weight management, not for sleep apnea as such.^[77] There, apnea counts only as a weight-related condition when deciding on weight treatment.^[77]

Restless legs

Restless legs syndrome is an urge to move the legs, often with an itch-like feeling, that is worse at rest and in the evening and eases when you move. The AASM guideline (published 2024, print 2025) starts with the things that make it worse: alcohol, caffeine, some antihistamines and antidepressants, and untreated sleep apnea.^[37] It also starts with iron, measured with two blood tests. Ferritin is a marker of the body's iron stores, and transferrin saturation is a measure of how much iron the blood is carrying.

For people with low iron stores, the guideline strongly recommends one iron product given through a vein (IV), ferric carboxymaltose.^[37] Other IV forms and iron tablets get weaker, conditional recommendations.^[37] The table shows which iron the guideline considers at which test result. These cut-offs are expert consensus, not tested thresholds.^[37]

Source: AASM guideline.^[37]

Among medicines, gabapentin enacarbil, gabapentin and pregabalin get strong recommendations.^[37] Dopamine drugs such as pramipexole and ropinirole were the old first choice. They are now advised against for standard long-term use (a conditional recommendation), because symptoms tend to get worse over time on them.^[37] Ask your doctor which of these are licensed in Germany. Periodic limb movements in sleep, leg twitches that show up on a sleep study, are treated only if they disturb you.^[37] The guideline makes few recommendations for that condition.^[37]

Acting out your dreams (RBD)

REM sleep behavior disorder (RBD) is the most urgent red flag in this guide. In RBD, the muscle paralysis that normally comes with dreaming fails, so people kick, punch, shout or leap out of bed during dreams. A multicentre study followed 1,280 people whose RBD had been confirmed in a sleep lab and who had no parkinsonism or dementia at the start.^[36] Most were men in their sixties.^[36] About 6.3 percent a year developed Parkinson's, Lewy body dementia or multiple system atrophy, and the estimated risk after 12 years was 73.5 percent.^[36]

That makes RBD a strong early warning sign for this family of brain diseases. The referral therefore goes to neurology, as much for monitoring and a possible place in trials of brain-protective treatments as for symptom relief. Doctors may use immediate-release melatonin, often starting at 3 mg and raised step by step up to about 15 mg, or low-dose clonazepam (0.25 to 1 mg).^[71] Clonazepam can cause drowsiness, falls and memory problems and can worsen sleep apnea.^[71]

Make the bedroom safe right away. Remove sharp or heavy objects and anything that could be used as a weapon, pad or move furniture away from the bed, and put a mat beside the bed.^[71] Until it is treated, consider sleeping apart from your partner.^[71]

Body-clock disorders and narcolepsy

Delayed sleep-wake phase disorder, the clinical form of being a night owl, is a diagnosed condition, so see a sleep clinician before you start treating it. The usual approach combines bright morning light with a low dose of melatonin, taken several hours before your natural evening melatonin rise. A sleep clinician should set that timing, and low doses of 0.5 to 1 mg are often enough to shift the clock.^[82] The early-bird version, shift-work disorder and the non-24-hour rhythm in fully blind people each have their own protocols.

Narcolepsy type 1 is marked by a lack of orexin, the brain's stay-awake signal. It usually also brings cataplexy, sudden muscle weakness set off by strong emotion, although cataplexy can be missing. Type 2 has no cataplexy. Diagnosis needs an overnight study plus a daytime nap test that checks how fast you fall asleep and whether you go straight into REM.

Where to go in Germany, Austria and Switzerland

Each of the three countries has its own route to a sleep specialist:

• In Germany, the DGSM (Deutsche Gesellschaft für Schlafforschung und Schlafmedizin) keeps the list of accredited sleep centres. Statutory insurance (GKV) covers the overnight study through a stepwise process (as of October 2026). It also reimburses CPAP machines and supplies as medical aids, with your usage checked over time. Insomnia apps listed in the DiGA directory can be prescribed on statutory insurance.
• In Austria, the ÖGSM is the sleep-medicine society, with sleep labs at AKH Wien, Graz and Innsbruck.
• In Switzerland, the SGSSC oversees the Fähigkeitsausweis Schlafmedizin, the formal sleep-medicine qualification, so it is the title to look for in a specialist.

Whichever country you live in, the red flag that applies to you sets the first step. A two-week sleep diary then gives the doctor your actual nights to work with.

Consumer sleep trackers are good at total sleep time and trends, rough at sleep stages and poor at spotting time awake. The reference is a lab sleep study (polysomnography, or PSG). Even there, a trained human scorer matches the expert consensus only about 86 percent of the time.^[39] That disagreement among experts adds some uncertainty to every comparison of a device with the lab.

Measured against that reference, wrist trackers do well on total sleep and less well on stages. A 2025 study tested six wrist devices against the lab in 62 adults.^[47] The table shows how far five of them were off on total sleep and how well they agreed with the lab on sleep stages, where 1.0 means perfect agreement.

Source: Schyvens et al., Sleep Advances 2025.^[47]

The weak spot behind those numbers is wake time. The devices caught more than 90 percent of the time people were asleep but only about 29 to 52 percent of the time they were awake.^[47] So they tend to score restless time as sleep.^{[38, 47]} If your nights are broken or you have insomnia, your tracker probably makes them look better than they are. Deep sleep is a guess as well. Wrist devices correctly labelled only about 47 to 70 percent of lab-scored deep sleep,^[47] because they infer it from heart rate and movement rather than measuring brainwaves.

The Oura Gen3 ring did well in a study that Oura partly funded, with about 92 percent accuracy for asleep versus awake.^[40] Its stage accuracy was about 76 to 91 percent, depending on the stage.^[40] Because that study used different measures and was partly funded by Oura, its numbers do not line up with the table above.

Only headbands that record brainwaves (EEG) measure sleep stages directly. The Dreem headband matched lab scoring of deep sleep about as well as two human experts agree with each other.^[39] The Muse S Athena (about $475, as of October 2026) is also a forehead EEG headband, but it has no independent published validation of its sleep staging.

Apnea screening is the one area where a consumer device has a clinical comparison. In people with suspected apnea, the Withings Sleep Analyzer, a pad that lies under the mattress, picked up 88 percent of those with moderate or worse apnea.^[41] It correctly ruled apnea out in about 89 percent of those without.^[41] That study was run largely by Withings staff, so treat it as promising rather than settled. Recent Apple Watch models can also flag possible sleep apnea from breathing disturbances at night. Either result is a nudge to get tested rather than a diagnosis. A proper take-home or lab test is needed both for the diagnosis and for insurance to cover treatment.

What to choose follows from those results:

• For total sleep time and trends: any modern wearable, with the Apple Watch closest on total sleep in the 2025 study.^[47]
• For deep sleep: an EEG headband such as the Dreem, which showed lab-level agreement in its validation study.^[39]
• For apnea screening: the Withings Sleep Analyzer (about €150 at Withings, as of October 2026), followed by a clinic if it flags apnea.^[41]

Whatever you wear, read it as a trend across weeks rather than a verdict on one night.

Sleep supplements can help you fall asleep faster and feel more rested, and that is where their best results show up, along with anxiety and mood, rather than in deep sleep. Melatonin leads the shelf: of the common options, it is the only one with an approved EU health claim for sleep. Matching each product to what it has actually been shown to do is how you get the most out of this shelf, and our deep sleep guide covers deep sleep itself in detail.

Melatonin

Melatonin has two approved EU health claims. It helps relieve the subjective feeling of jet lag at 0.5 mg or more, taken close to bedtime on the travel day. It also helps you fall asleep faster at 1 mg, taken close to bedtime. The effect is real and modest. In a meta-analysis, melatonin cut the time to fall asleep by about 7 minutes and added about 8 minutes of sleep, with a small gain in sleep quality.^[42] Higher doses and longer trials gave somewhat bigger effects on falling asleep and total sleep, but not on sleep quality.^[42] For jet lag, the CDC advises against doses above 5 mg,^[82] and high doses can leave you groggy the next morning.

Long-term safety is the open question. In August 2024, Germany's Federal Institute for Risk Assessment (BfR) warned that the risks of melatonin supplements, especially with long-term use, are poorly studied. It also said it cannot set a safe daily dose. It advises against uncontrolled use, especially for children and teenagers and during pregnancy, breastfeeding or when trying to conceive. The same applies to people with liver or kidney problems, autoimmune disease or epilepsy. People at risk of type 2 diabetes should get medical advice first. Even doses of 1 mg or less can cause drowsiness, headache and next-day effects on driving.^[75]

Heart safety is a second open question. A 2025 analysis of US health records, presented at a conference, reported more heart failure in people with insomnia who used melatonin long term.^[76] It was observational, though, and could not rule out other explanations.^[76] A 2026 analysis of US survey data found no clear link between recent melatonin use and existing heart failure.^[76] It did not look at years of use, though, so it cannot settle the long-term question. In Germany, melatonin medicines are prescription-only.^[75]

Small effects on falling asleep and how sleep feels

Beyond melatonin, a few other supplements show promising results in small trials, mostly for falling asleep and for how rested people feel. In a small trial in healthy adults, 200 mg of L-theanine (an amino acid from tea) a day for 4 weeks shortened the time to fall asleep compared with placebo.^[43] It also reduced sleep disturbance and the use of sleep medication.^[43] The overall sleep-quality score improved too, but it did not clearly beat placebo.^[43] Saffron extract (affron) improved self-rated sleep quality compared with placebo in two 28-day trials in adults who slept poorly.^{[65, 66]} The doses were 14 mg twice a day or 14 to 28 mg in the evening.

Magnesium is an easy option to try if you are slow to fall asleep. A meta-analysis of 3 small trials in older adults found that it cut the time to fall asleep by about 17 minutes, but the evidence quality was low.^[44] To try it well, stay at or below 250 mg a day of magnesium from supplements, the tolerable upper level set by EU food-safety scientists.^[83] Above it, diarrhoea becomes more likely.^[83] If you have kidney disease, ask your doctor first.

None of those magnesium trials tested magnesium bisglycinate, the form often sold for sleep. Magnesium does have approved EU health claims, for example that it "contributes to a reduction of tiredness and fatigue", but none of them is about sleep. Glycine is another simple candidate: 3 g before bed shortened the time to reach deep sleep and made sleep feel better in a small trial.^[45] It did not add any deep sleep, though.^[45] Like L-theanine and saffron, glycine has no approved EU health claim for sleep.

What to skip

Some popular options have less behind them than their labels suggest:

• Taurine on its own: we found no human trial that tested it with sleep as the main outcome.
• Multi-ingredient "sleep formulas": they often combine many ingredients at small doses, usually below the amounts tested in the trials above.
• Over-the-counter antihistamine sleep aids (diphenhydramine, doxylamine): they blunt thinking and memory, especially in older people, so older adults should avoid them.

Whatever the label says, when a product promises deeper sleep, ask how that was measured: with a lab brainwave recording, with a wearable's estimate, or only with a questionnaire. Many deep-sleep claims rest on the last two.

Most common sleep complaints have a first step you can take tonight and a point at which a doctor should take over. Each answer below links to the section with the evidence.

"I can't fall asleep"

Trouble falling asleep usually starts with the basics: a cool bedroom, dim light in the evening,^[13] and a fixed wake time. Have your last caffeine about 8 hours before bed, or about 10 if you are sensitive^[19] and no alcohol in the last 3 hours. If the problem has lasted 3 months or more, CBT-I is the treatment, with sleep restriction and stimulus control at its core.^[30] The CBT-I section explains how it works. If it started within weeks of a new medication, suspect the medication and talk to your doctor.

"I wake at 3 AM and can't get back to sleep"

Waking in the second half of the night usually has a cause worth finding. Alcohol breaks up sleep as it leaves your blood.^[20] Sleep apnea is likely if you also snore, someone has seen you stop breathing, or you wake with headaches. Depression or anxiety can show up as early waking, and so can needing to pee or hot flushes around the menopause. Treat the cause rather than the wake-up itself, starting with the red flags in the medical section.

"I sleep 8 hours but wake up exhausted"

Waking up exhausted after a full night is often a sleep-quality problem, such as undiagnosed apnea, sleep broken by alcohol or late meals, or depression. Persistent tiredness can also come from other medical causes, for example anaemia or thyroid problems, so see a doctor if it lasts. A wearable can show trends but cannot diagnose this.

"I'm a night owl and can't function in the morning"

A delayed body clock is a likely reason why a night owl struggles in the morning. Because it is a diagnosable condition, a sleep clinician can confirm it and supervise the fix. That fix pairs morning outdoor light with a low dose of melatonin taken several hours before your natural evening melatonin rise, as the medical section explains. If you are a night owl by nature, building your schedule around it can be the better choice, since social jetlag is linked to a higher BMI.^[52]

"I get jet lag for a week"

Jet lag is usually harder after flying east than after flying west, because your body clock moves later more easily than earlier.^[14] The best timing for light depends on the direction and on how many time zones you cross. For big shifts the simple "morning light going east" rule can backfire, so use our free jet lag calculator. Low-dose melatonin (0.5 to 1 mg is often enough) taken before bedtime at your destination can help when flying east.^[82]

"I snore loudly and my partner complains"

Loud snoring is worth screening for apnea with the STOP-BANG checklist from the medical section, and if you score 3 or more, ask for a take-home or lab test. Snoring on its own is not apnea, but loud snoring together with daytime sleepiness and pauses in breathing that someone has seen makes it likely.

"I wake up screaming or thrashing"

Screaming, kicking or thrashing while dreaming points to REM sleep behavior disorder and needs a neurology referral. Make the bedroom safe straight away by removing sharp or heavy objects and padding or moving furniture.^[71] Sleepwalking and night terrors are different, because they come out of deep sleep rather than dreams. They are usually set off by sleep loss, alcohol or fever, so treat the trigger and make the bedroom safe as well.

"My legs feel weird and I can't stop moving them"

An urge to move your legs that is worse at rest and in the evening and eases when you move is the restless legs pattern. Ask your doctor to check ferritin and transferrin saturation, the two blood tests for iron, because iron is one of the first things the guideline looks at.^[37] The treatment options are in the medical section.

"I take sleeping pills every night and want to stop"

Never stop benzodiazepines or Z-drugs abruptly after regular use, because with benzodiazepines withdrawal can be severe and, rarely, life-threatening.^[73] Plan a slow taper with your doctor, using the steps in the CBT-I section, and run CBT-I alongside it.^[54] For many people the taper takes months.^[73]