How to Get More Deep Sleep (What the Trials Actually Show)

Your watch says you got 41 minutes of deep sleep and you want the number higher. Almost every article about how to do that gives the same list: dark room, no screens, go to bed earlier. Those are fine, and they are mostly advice about falling asleep rather than advice about deep sleep, which is a specific stage with its own rules.

This page is about the things that have actually been tested in trials, what each one is worth, and the part almost nobody mentions: your tracker is guessing at that number, and it guesses at deep sleep worse than at anything else it reports.

Quick Answer

The four things with real trial evidence behind them: keep the same sleep and wake time every day, exercise regularly (evening sessions are fine, just not hard ones in the last hour), take a warm bath or shower one to two hours before bed, and stop drinking alcohol in the evening. Deep sleep is front-loaded into the first half of the night, so going to bed earlier and being woken less protects it. Most adults get 13 to 23 per cent of the night in deep sleep and it falls naturally with age. Your tracker’s deep sleep figure is an estimate, and the research puts its stage-by-stage accuracy near 50 per cent.

In this guide:

• What Deep Sleep Actually Is
• How Much You Should Get
• What Actually Increases It, Ranked
• 1. Same Time Every Day
• 2. Exercise, Including in the Evening
• 3. The Warm Bath, and Why the Timing Matters
• 4. The Alcohol Trap
• 5. Caffeine, Earlier Than You Think
• Your Tracker Is Guessing
• What Does Not Work
• When Low Deep Sleep Needs a Doctor
• The Verdict

What Deep Sleep Actually Is

Deep sleep is stage N3, also called slow-wave sleep because of the large, slow electrical waves that dominate the brain during it. It is the hardest stage to wake someone from: pulled out of it, people are groggy and confused for several minutes, which is the feeling of being woken by an alarm at the wrong moment.

Two features of it decide everything else on this page.

• It is front-loaded. Slow-wave sleep is concentrated in the first two or three cycles of the night and thins out towards morning, while REM does the opposite. That single fact means the hours before midnight are not magic, but the first hours of your own sleep are: anything that delays or interrupts them costs deep sleep specifically.
• It is pressure-driven. The longer you have been awake, the more slow-wave sleep your brain takes when it finally gets the chance. This is why a short night is followed by a night with more deep sleep than usual, and why you cannot simply add deep sleep on top of an already well-slept night. Most of what follows works by protecting it, not by manufacturing more.

Why it matters: deep sleep is when growth hormone is released, when the immune system does much of its work, and when the brain appears to clear metabolic waste. A 2023 study in JAMA Neurology followed adults over 60 and found that each percentage point of slow-wave sleep lost per year was associated with a higher risk of developing dementia. That is an association in older adults, not proof that raising your number protects you, and it should be read that way. It is still the most serious reason to care about this stage rather than just collecting the metric.

If the word “core” on your watch is what brought you here, that is a different label for a different thing, and it is explained in what is core sleep.

How Much You Should Get

For a healthy adult, deep sleep is roughly 13 to 23 per cent of total sleep time. On a seven-and-a-half hour night that is about 60 to 105 minutes.

Age is the biggest single factor and you cannot train it away. Slow-wave sleep peaks in childhood, falls steeply through the teens and twenties, and keeps declining after that. A man in his sixties may get half the deep sleep of the same man at twenty-five, with no illness involved. Comparing your number to a 22-year-old’s, or to an app’s generic target, produces anxiety rather than information. Compare it to your own average over a month.

The more useful target is total sleep. Deep sleep is a proportion of the night, so the fastest way to get more of it is to have more night. Going from six hours to seven adds roughly ten minutes of deep sleep before you change anything else. Our sleep debt calculator shows how far the last two weeks have put you behind.

What Actually Increases It, Ranked

Ranked by how good the evidence is and how much it is likely to change, not by how often the advice is repeated.

1. Same Time Every Day

This is first because of how strong the evidence has become, and because it is the one most people ignore while doing everything else on the list.

A 2024 analysis in Sleep of more than 60,000 UK Biobank participants wearing accelerometers found that how regular someone’s sleep was predicted mortality risk better than how much sleep they got. The most regular sleepers had substantially lower risk than the most irregular, and regularity outperformed duration as a predictor. A separate UK Biobank analysis in eLife the same year reached the same conclusion.

Why it affects deep sleep specifically: the pressure that produces slow-wave sleep builds on a roughly 24-hour rhythm, and that rhythm is set by when you consistently sleep and when you see light. A weekend that shifts your sleep by two hours puts the deep-sleep-heavy first cycles at a time your body clock is not ready for them, and you lose part of the stage to lighter sleep and awakenings.

• Fix the wake time first. It is the anchor, it is under your control, and bedtime follows it within about a week.
• Keep the weekend within an hour of the weekday. Not identical, within an hour.
• Get daylight soon after waking. Ten minutes outdoors does more for the clock than any indoor light.

Key Takeaways

• Deep sleep is front-loaded into the first cycles, so protecting the start of the night matters more than anything you do at 5am.
• Regularity beats duration: sleep timing consistency predicted mortality better than sleep length in 60,000 people.
• Evening exercise does not wreck sleep. The meta-analysis says it slightly helps, unless it is vigorous and ends within an hour of bed.
• A warm bath works by dumping heat afterwards, which is why the timing is one to two hours before bed and not right before.
• Alcohol raises deep sleep in the first half and takes it back with interest in the second. The night nets out worse.
• 13 to 23 per cent of the night is normal, it falls with age, and your watch’s figure for it is an estimate with roughly coin-flip stage accuracy.

2. Exercise, Including in the Evening

A meta-analytic review in the Journal of Behavioral Medicine pooled trials of physical activity and sleep and found regular exercise improved sleep quality and increased slow-wave sleep, with modest effect sizes. Modest is the honest word: exercise is not going to double the number. It is one of the few interventions that reliably moves it at all.

The evening exercise rule you have been given is mostly wrong. A systematic review in Sports Medicine examined evening exercise in healthy people and found that, on average, it did not harm sleep and slightly increased slow-wave sleep. The exception was clear and narrow: vigorous exercise ending less than about an hour before bed did delay sleep onset and reduce sleep efficiency. A network meta-analysis in Nature and Science of Sleep reached a similar conclusion about intensity being the variable that matters.

• Train whenever you will actually train. A consistent evening habit beats a morning plan you skip.
• Leave at least an hour, ideally two, between a hard session and bed. Easy sessions need no gap.
• Aerobic and resistance training both count. Neither is clearly better for slow-wave sleep.
• The effect builds over weeks. One workout does not buy one good night; a month of training changes the baseline. If you want a structure for that, the FITT principle is the four dials to set.

3. The Warm Bath, and Why the Timing Matters

This one sounds like a spa tip and is actually one of the better-evidenced items on the page. A systematic review and meta-analysis in Sleep Medicine Reviews pooled 13 studies of passive body heating by warm bath or shower and found it improved sleep quality and shortened the time taken to fall asleep, with the effect strongest when the bathing happened one to two hours before bedtime at around 40 to 43°C for 10 minutes or more.

The mechanism explains the timing, and the timing is the whole trick. Falling asleep requires your core body temperature to drop. A warm bath pushes blood to the skin of your hands and feet, and when you get out, that blood radiates heat away and your core temperature falls faster and further than it would have. You are not warming yourself to sleep; you are setting up a bigger cooling slope. Do it five minutes before bed and you get the warming without the cooling, which is the opposite of what you want.

A shower works, the water should be warm rather than hot enough to be uncomfortable, and if neither appeals, a foot bath uses the same mechanism through the same blood vessels.

4. The Alcohol Trap

Alcohol is the reason a lot of people think they need help with deep sleep, and the reason their tracker sometimes tells them the opposite.

A drink before bed increases slow-wave sleep in the first half of the night. That is a real finding, repeatedly measured, and it is why a nightcap feels like it works. The review of alcohol and the sleeping brain in the Handbook of Clinical Neurology sets out what happens next: as the alcohol is metabolised, the second half of the night is fragmented, REM rebounds, awakenings increase, and sleep quality falls. Polysomnography work on acute dosing found the same pattern of suppressed REM and disrupted architecture. The night nets out worse, and a tracker that reports a good deep sleep figure is reporting the first half only.

• Three hours between the last drink and bed is a workable rule for one or two drinks. More than that needs more time.
• The effect is dose-dependent. One glass with dinner is not the same as three after it.
• If you want to see it for yourself, compare a two-week stretch with no evening alcohol against your normal fortnight. This is the single change most likely to produce a visible difference in how mornings feel.

5. Caffeine, Earlier Than You Think

A 2025 meta-analysis of controlled crossover trials in Sleep Medicine looked at caffeine’s effects by age and dose and confirmed the pattern: caffeine lengthens the time to fall asleep, reduces total sleep time and reduces slow-wave sleep, with larger doses and older participants affected more.

The practical problem is that caffeine’s half-life is about five to six hours, so a 3pm coffee still has a quarter of its dose working at midnight. That dose lands exactly on the first cycles, which is where your deep sleep lives.

• Cut off 8 to 10 hours before bed, not the usual six. For an 11pm bedtime that is early afternoon.
• Half-life varies a lot between people, by genetics, smoking, pregnancy and some medications. Our caffeine half-life calculator shows how much is still in your system at bedtime from a given cup and time.
• “It does not affect me” is usually false. People who sleep fine after evening coffee still show measurably less slow-wave sleep in the lab. You do not feel the stage; you feel the falling asleep.

Your Tracker Is Guessing

This section is here because the number you are trying to improve may not be the thing you think it is.

Sleep stages are defined by brain electrical activity. A watch or a ring cannot see that. It infers stages from movement, heart rate and heart rate variability, and the inference is hardest exactly where you are looking. A 2024 systematic review in JMIR mHealth and uHealth compared Fitbit, Garmin and WHOOP against polysomnography and found that while they are reasonably good at telling sleep from wake, their agreement with the lab on individual stages is far weaker, with deep sleep among the least reliable. A multicentre validation in Sensors the same year tested three commercial devices and reported the same ordering.

What that means in practice:

• One night’s deep sleep figure is close to meaningless. The error on a single night is large enough to swamp any change you made that day.
• The trend over a month is worth something. Errors that are roughly consistent still let you see direction, which is all you need.
• Do not compare brands. Two devices on the same person on the same night routinely disagree by a large margin, because each uses a different algorithm.
• If the number is making you anxious, stop looking at it. Worrying about sleep measurements is a documented way to sleep worse. The best evidence for any of the advice above is how you feel at 3pm, not what the app says at 7am.

What Does Not Work

• Melatonin, for this purpose. It is useful for shifting when you sleep, for jet lag and for delayed sleep phase. It is not a deep sleep drug and the trials do not show it increasing slow-wave sleep.
• Magnesium. Popular, cheap, and the evidence for it improving sleep architecture in people who are not deficient is weak and inconsistent. If you take it and sleep better, there is no harm in continuing, but do not expect the deep sleep number to move.
• Sleeping longer on weekends. Catch-up sleep recovers some of the deficit but breaks the regularity that the mortality data cares about. It is a patch, not a plan.
• Buying a better tracker. A more expensive estimate is still an estimate.
• Alcohol, however it feels. Covered above, and it is the most common self-prescribed sleep aid there is.
• Chasing a number that age has changed. If you are 55 and getting less deep sleep than you did at 25, that is biology, not a failure of routine.

When Low Deep Sleep Needs a Doctor

Most low numbers are lifestyle, age, or the tracker. Some are not, and the difference is usually daytime.

• Loud snoring, gasping or witnessed pauses in breathing. Sleep apnoea fragments the night and strips slow-wave sleep, and it is both common and treatable. This is the single most important thing on the page to rule out.
• Sleeping a full night and still being exhausted all day, for weeks.
• Falling asleep without meaning to, especially while driving.
• Restless, crawling sensations in the legs at night that force you to move them.
• Sleepwalking, night terrors or confusional waking in an adult. These arise out of deep sleep and are worth a conversation.
• Persistent insomnia. The first-line treatment is CBT for insomnia, not a sleeping pill, and most sedatives suppress slow-wave sleep rather than increasing it.

The Verdict

Deep sleep is front-loaded, pressure-driven and largely set by age, which means you cannot manufacture much of it. What you can do is stop losing it. Keep your sleep and wake times steady, sleep long enough that the proportion has something to work with, exercise regularly and do not finish hard sessions on top of bedtime, take a warm bath one to two hours before bed, keep alcohol out of the evening, and move your caffeine cut-off earlier than feels necessary.

That is the whole evidence-based list. Everything else sold for this is either aimed at falling asleep rather than at this stage, or aimed at your wallet.

And hold the number loosely. Your watch is estimating a brain state it cannot see, so judge the changes by how the afternoons feel over a month, not by one morning’s chart. For what that chart is actually showing you, see what is core sleep, and for how far behind the last fortnight has put you, the sleep debt calculator.