Deep sleep (slow-wave sleep)

Deep sleep, slow-wave sleep and N3 are three names for the same stage of non-dream sleep. Sleep doctors call it N3, and on a brain-wave readout it shows up as slow-wave sleep (SWS). "Deep sleep" is the friendlier word your watch puts on the screen, although the watch only estimates it.

In a sleep lab, those slow waves define the stage: a 30-second window counts as N3 when big, slow waves fill at least 20 percent of it.^[72] Each wave is a huge crowd of brain cells switching on and off in sync. Two faster rhythms ride on top of them: sleep spindles, short bursts of quicker brain activity, and ripples, very fast bursts from the brain's memory hub (the hippocampus). When the three line up, the brain files away the day's memories.^{[8, 9]}

That filing happens early, because deep sleep is not spread evenly across the night. In healthy young men aged 16 to 25, N3 filled about a fifth of the night, about 19 percent.^[1] Most of it comes in the first sleep cycles, while dream sleep (REM) dominates the last ones,^[9] which is why an early alarm mostly takes dream sleep.

A late bedtime works the other way round, because the extra time awake builds sleep pressure. In a home-recording study, going to bed about three hours late raised the share of deep sleep while the share of dream sleep fell.^[79] Older men showed less of this rebound.^[79] A short night costs you sleep either way; the body simply protects deep sleep first.

Protecting deep sleep first does not make it the only stage that counts. N2, the lighter middle gear, is the biggest single slice of the night, and its spindles help with memory too.^[8] Treating N3 as the only "good" sleep misses that every stage of sleep does its own work.

N3 still gets the longevity attention, for three reasons. In men, the biggest growth-hormone pulse of the day comes at night, mostly in that first deep stretch, and the amount tracks deep sleep.^[1] The fluid around the brain pulses in step with deep-sleep waves.^[6] And N3 fades first with age: by midlife, the men in the key study had a small fraction of the deep sleep they had at 20.^[1]

Deep sleep matters for three jobs researchers can trace: releasing growth hormone, moving fluid through the brain and filing memories. Each has solid evidence for a link with N3, and none proves that more deep sleep means a longer life.

Growth hormone has the clearest link. It is the same hormone that makes children grow; in adults, whose growth plates have closed, it supports tissue repair, fat use, IGF-1 (a growth signal) and bone renewal. In healthy men, deep sleep and growth-hormone secretion both fell sharply over the same years, between young adulthood and midlife.^[1] A small study in young men points the same way: a drug (gamma-hydroxybutyrate) raised deep sleep, and growth hormone rose in step with it.^[2]

The brain's overnight rinse is the second job, and its evidence is more mixed than the headlines. In mice, researchers described a network that pumps fluid through the brain and carries away waste.^[44] A later mouse study found that the spaces between brain cells widen during sleep, which sped up the clearance of amyloid-beta, the protein that clumps in Alzheimer's.^[5] In humans, a brain-imaging study of 13 sleeping adults saw the slow waves of non-dream sleep followed by waves of blood flow.^[6] Pulses of cerebrospinal fluid, the fluid that bathes the brain, came next.^[6]

That human study did not measure waste removal, though. Mouse studies disagree on whether clearance rises.^[5] or falls^[7] during sleep, so whether deep sleep washes waste out of the human brain is still open.

Memory, the third job, has the best-worked-out model. A rat study first showed that the hippocampus replays the day's experiences during sleep^[39] Two large reviews describe how that replay is timed to spindles and slow waves, slowly moving memories into long-term storage.^{[8, 9]} Lose deep sleep, and recall for facts suffers.^[8]

Weaker memory filing is also where the dementia question comes in. A review tied age-related deep-sleep loss to thinning in the front of the brain, fewer spindles and poorer memory filing.^[4] The Whitehall II study then looked at sleep length in almost 8,000 British civil servants.^[12] Sleeping 6 hours or less at age 50 was linked to about a 22 percent higher risk of dementia later.^[12] The risk was higher still in people who stayed short sleepers through their 50s, 60s and 70s.^[12] That study measured only how long people slept, not deep sleep, so lost N3 is one possible path from short nights to dementia, not a proven one.

None of this makes deep sleep a stand-alone target. A long night with less N3 still gives you every other stage, while a short night with a high N3 share does not. Deep sleep does important work, but no study has shown that pushing your share higher makes you live longer, so never trade away total sleep to bump the percentage.

Deep sleep is the stage that disappears earliest. In healthy men, it fell from about 19 percent of the night at ages 16 to 25 to about 3.4 percent at 36 to 50, and then barely changed.^[1] The table turns those shares into minutes on a 7.5-hour night:

Source: Van Cauter et al..^[1] These are strict sleep-lab scores, so a wearable may show quite different numbers.

The numbers come from an analysis that pooled sleep-lab recordings of 149 healthy men aged 16 to 83 from four laboratories.^[1] Most of the loss happened between young adulthood and midlife. After that the deep-sleep curve flattened, while time awake at night kept rising by about 28 minutes per decade.^[1] Dream sleep held fairly steady until midlife and only then fell, by about 10 minutes per decade after 50, so in these men the deep-sleep loss came decades earlier.^[1] The study compared age groups rather than following the same men over time, and it included no women.

A larger meta-analysis of 65 studies in healthy people aged 5 to 102 confirmed the direction: the deep-sleep share falls with age in adults.^[3] How big the age effect looked, though, depended heavily on how carefully the studies had screened out illness, medicines and sleep apnea.^[3]

Those three can speed the decline up, which is a good reason to rule them out before blaming age alone. Obstructive sleep apnea (the airway repeatedly collapses during sleep) fragments the night and cuts into deep sleep, and chronic insomnia and depression disturb sleep as well. Some medicines change how sleep is built across the night: benzodiazepines in particular reduce deep sleep. Ask your doctor before changing any prescription, because treating the underlying cause is often what lets deep sleep recover.

What remains once those causes are ruled out is mostly structural. A review links the decline largely to thinning grey matter in the front of the brain, the very region that generates slow waves.^[4] Fewer spindles, an earlier body clock and more waking at night add to it.^[4] Together they produce the familiar older-adult pattern of shorter nights, less deep sleep and more light sleep.^[4]

Older brains make fewer slow waves to begin with, so no habit or product will rebuild a 25-year-old's night. The realistic goal is to hold steady, or to win back a little by clearing away what still disturbs your sleep.

Rings and watches only estimate deep sleep. In an independent lab test, wrist devices caught roughly half to two-thirds of it, so their nightly number is a rough trend at best.^[43] A forehead band that reads real brain waves gets closer,^[54] and in older adults the gap between wearable and lab is larger still.^[55]

The benchmark is a full sleep-lab study (polysomnography), where trained experts score deep sleep from brain waves by the official rules. Even those experts agreed with a consensus score only about 86 percent of the time, window by window.^[35] So experts disagree on about one window in seven, and no device should be expected to beat that agreement by much.^[35]

Against that yardstick, an independent test of six wrist devices in 62 adults, one lab night each, measured how much of the lab-scored deep sleep each device caught.^[43]:

Source: Schyvens et al. 2025^[43] The sixth device, the Withings Scanwatch, lumps deep and dream sleep together, so it is not comparable; the Apple Watch did best across all stages combined.

Rings show a split picture. In an Oura-funded study of 96 Japanese adults, the Oura Gen3's average deep-sleep time matched the lab's at group level.^[36] A small independent lab comparison in adults aged 56 to 80 found the opposite.^[55] There, the Oura ring overestimated deep sleep by about 70 minutes and two bedside sensors by up to about an hour and a half.^[55] A Fitbit watch went the other way, underestimating it by about half an hour, and all of them staged deep sleep poorly.^[55]

Both findings can hold at once, because a group average that matches the lab can still hide large errors on single nights when overestimates and underestimates cancel out. Wearables cannot see brain waves; they infer deep sleep from heart-rate patterns and movement. So treat deep-sleep minutes on any wearable as a rough trend, best read as a 7-day rolling average over several weeks, especially past 55.

For real brain-wave data at home, the best-tested consumer option is Muse's forehead headband. In an academic lab study of 47 adults, the Muse-S agreed well with the sleep lab, both overall and on deep sleep.^[54] It did read about 15 minutes more deep sleep than the lab.^[54] That study tested the Muse-S; the newer Muse S Athena has not been validated separately in a published study. Price: about 475 US dollars on the maker's site (as of October 2026).

Dreem headbands were the research-grade alternative. In a company-run study, the Dreem's automatic scoring matched the experts 83.5 percent of the time, close to the experts' own 86.4 percent.^[35] Dreem headbands are no longer sold to consumers, so a used one is only worth anything if its app and data access still work.

Measuring deep sleep is one thing; some headbands also promise to raise it with sound. They rest on a real lab method, closed-loop auditory stimulation (CLAS), which plays short pulses of pink noise timed to the brain's own slow waves. In small lab studies of young,^[10] and older adults^[11] the pulses raised slow-wave activity while they played. In the older adults, though, the whole-night total did not differ from a sham session without the pulses.^[11] Both studies also reported better memory the next morning, which is promising but far from settled in groups that small.^{[10, 11]}

The consumer version, Philips' SmartSleep headband, is no longer sold, and as of October 2026 no FDA-cleared, peer-reviewed consumer CLAS device exists. Some consumer headbands now advertise sound-based "deep sleep boost" features backed only by company data. Real CLAS has to read your brain waves live and time each pulse to them, so a device that loops pink noise on a timer is not doing what the studies tested.

If deep sleep is what you want to track, three rules cover the buying decision:

• Best brain-wave option: a Muse headband, read as a trend.^[54]
• Best ring or wrist estimate: an Oura Gen3 or one of the better-tested watches, read as a 7-day average and never as a one-night verdict.^{[36, 43]}
• Skip "deep sleep enhancement" gadgets that have no published sleep-lab validation.

The habits with real evidence for deep sleep are evening exercise, treating sleep apnea and keeping enough sleep pressure. Regular sleep times and a calm evening support all three.

Evening exercise has a measured but small effect. A meta-analysis of 23 studies found that a single evening workout raised deep sleep by about 1.3 percentage points compared with no exercise.^[18] On a 7.5-hour night, that works out to roughly 6 minutes. Vigorous exercise that ends within an hour of bedtime may make it harder to fall asleep, though.^[18] Why exercise helps is not settled: in that analysis, a higher body temperature at bedtime went with slightly worse sleep, not better.^[18]

A reasonable routine follows from those limits: 3 to 4 moderate to hard sessions a week, with the hard ones finished at least an hour before bed. The measured effect comes from single sessions, so the weekly number is a sensible habit rather than a tested dose.

Untreated sleep apnea can cancel that gain and more. Repeated breathing pauses fragment sleep and cut into deep sleep, and treatment often brings deep sleep back. The most common treatment is CPAP, a mask that keeps the airway open; for mild to moderate apnea, a custom oral appliance is an option.^[57] If you have moderate-to-severe apnea and don't know it, the habits in this guide cannot make up for it.

Restless legs syndrome breaks up sleep in a similar way, and low iron is one treatable cause. A doctor can check your iron stores (ferritin) before you take anything, since the AASM guideline ties iron treatment to blood-test cut-offs.^[42] The details are in the 30-day plan below.

Once nothing is breaking up the night, sleep pressure decides how deep it gets: the longer you are awake, the more intense your next deep sleep.^[78] Insomnia therapy (CBT-I, cognitive behavioural therapy for insomnia) uses this on purpose with sleep restriction, a structured treatment with a therapist or a validated programme. Cutting your own sleep short is not a way to get more deep sleep, because you only get a denser night with less total sleep. If sleep restriction sounds useful, do it with a clinician.

A steady body clock gives that sleep pressure a fixed time to land. Going to bed and getting up at the same times every day, weekends included, is the simplest way to keep the clock steady.

A racing mind at bedtime works against all of it. In insomnia, CBT-I targets that arousal. Its tools include reframing anxious thoughts and stimulus control, which means using the bed only for sleep and getting up when you cannot sleep. For everyone else, a calm wind-down with dim light, slow breathing or a warm bath is a reasonable way to quiet the system, and the bath brings its own evidence on temperature.

Temperature changes deep sleep mostly through your skin: gentle skin warming deepened sleep in a lab study, while a hot, humid bedroom cut deep sleep.^{[14, 16]} The mechanism starts with falling asleep. Your core temperature drops at night, and you shed that heat through your hands and feet, whose blood vessels widen to release warmth. In one study, how much warmer the hands and feet were than the torso predicted how fast people fell asleep.^[13] It did so better than core temperature, melatonin or sleepiness.^[13]

The strongest single result comes from a small sleep-lab experiment with a temperature-controlled suit (not, so far, standard bedroom wear). Warming the skin by just 0.4 degrees C almost doubled the share of deep sleep in older people and made early-morning waking much rarer.^[14] The study included young and older, healthy and insomniac sleepers, and the effect was strongest in the older ones. Because it used a special suit, not a bath or a room setting, it shows that skin temperature matters, not which household trick gets you there.

Room air turns out to matter less than people think, as long as it is not hot. A review found that with normal bedding, room air from about 13 to 23 degrees C did not change sleep stages.^[16] Humid heat, by contrast, cut deep sleep and caused more waking at 32 degrees C and 80 percent humidity.^[16] So the point is to avoid a hot bedroom. Around 17 to 20 degrees C is a common comfort recommendation, not a tested deep-sleep optimum. Bed socks are a cheap way to keep your feet warm if they run cold.

A warm bath before bed works on falling asleep rather than on proven extra deep sleep. A meta-analysis of 13 trials looked at a warm bath or shower of 40 to 42.5 degrees C, taken 1 to 2 hours before bed.^[15] Even 10 minutes shortened the time to fall asleep and improved self-rated sleep quality and sleep efficiency.^[15] The idea behind it is the same heat-shedding through hands and feet,^[13] and like any heat, it has limits.

Those limits start with the water temperature. Begin at about 40 degrees C, since 42.5 degrees C is the top of the studied range, not a target. Older adults and people with high blood pressure or heart disease should stay below 42 degrees C.^[52] Don't bathe after alcohol or right after a big meal, don't bathe alone if you are frail, and get out slowly so you don't get dizzy.^[52] If you have heart disease, are pregnant, or have low blood pressure, ask your doctor first; this also applies to sauna and to hard interval training.

Cooling mattresses have one solid trial behind them. In a trial at three centres, 72 people slept on one specific high-heat-capacity mattress and on a regular one.^[17] The special mattress gave about 7.5 extra minutes of deep sleep on average and a slightly lower heart rate, with large person-to-person variation.^[17] One co-author works for the mattress maker. No independent trial has shown the same for other cooling covers or toppers, so manufacturer-funded claims of bigger gains remain unverified.

Sauna has almost no sleep-lab evidence. In a large Finnish cohort, men who used the sauna often had lower cardiovascular death rates.^[77] That is an association, and it says nothing about sauna versus a hot bath for sleep. Like a bath, a sauna heats you up and is followed by a cool-down, so a sleep benefit is plausible but not shown.

In practice, the temperature steps come down to four:

• Room: not hot. With a duvet, anything from about 13 to 23 degrees C left sleep stages unchanged.^[16]; 17 to 20 degrees C is a common comfort pick.
• Feet: warm, with bed socks if they run cold.
• Bath or shower: about 40 degrees C for 10 minutes or more, 1 to 2 hours before lights-out, to fall asleep faster^[15] Stay below 42 degrees C if you are older or have high blood pressure or heart disease.^[52]
• Optional: a high-heat-capacity mattress, with about 7.5 extra minutes of deep sleep as the average to expect, not the marketing figure.^[17]

Supplements can help you fall asleep faster, take the edge off anxiety, make you rate your night higher or shift your body clock, and those are real wins. For deep sleep itself, one small magnesium trial in 12 older adults found a few more minutes,^[41] while convincing, repeated sleep-lab evidence is still missing for every supplement. Prescription and experimental drugs such as sodium oxybate and gaboxadol do raise deep sleep, but they are a different category.^{[2, 75]}

Most "deep sleep" product claims come from a wearable (Oura, Fitbit, Eight Sleep Pod) or a questionnaire, not from brain waves. Those numbers are not worthless, but they measure something else, so asking the seller for the sleep-lab study is a fair first question. Even for diagnosed insomnia, the European guideline does not recommend herbal products or fast-release melatonin as treatment.^[45]

Sorted by what human brain-wave evidence shows, sleep supplements fall into four groups.

The closest to a deep-sleep effect, from early, small studies

• Glycine, 3 g before bed: got people into deep sleep sooner, without adding more of it.^[19]
• Magnesium: one 12-person trial in older adults found a few more minutes of deep sleep.^[41]; the rest of the evidence is about falling asleep^[22]
• Lemon verbena plus zinc: verbena alone improved self-rated sleep in human trials, measured with a Fitbit rather than a sleep lab, while the deep-sleep claim for the combination is untested.^[24]

Best for faster sleep or a calmer evening

• L-theanine, 200 to 400 mg: calms without sedating and helps some people drift off.^{[29, 71]}
• Saffron extract, about 15 to 28 mg a day for 4 to 6 weeks: better self-rated sleep.^{[30, 65]}
• Melatonin at a low dose: shifts the body clock and shortens the time to fall asleep a little.^[31]; the melatonin details cover when it makes sense.

A plausible mechanism, no human sleep evidence

• Taurine on its own.
• Most multi-ingredient "sleep formulas", which spread many ingredients across small doses.

Better left out

• Diphenhydramine and doxylamine, the drugstore antihistamine sleep aids: the AGS Beers list says to avoid them past age 65^[63]
• Benzodiazepines, which tend to reduce deep sleep, and Z-drugs, whose effect on deep sleep is mixed; both bring tolerance, dependence and falls.^[53]
• Alcohol, which adds deep sleep early in the night, then breaks up the second half and cuts dream sleep.^[32]
• High-dose melatonin (5 to 10 mg): across trials, higher doses did not improve sleep quality, though the review did not compare high and low doses directly.^[31]

Glycine and magnesium are the best-supported sleep supplements, and both mainly help you fall asleep rather than add deep sleep.^{[19, 22]}

Glycine has the cleaner human data. In a small trial, people who slept badly took 3 g of glycine 30 minutes before bed.^[19] In the sleep lab, they fell asleep faster and reached deep sleep sooner than on placebo, and they rated their sleep better and felt less sleepy the next day.^[19] The overall shape of their sleep did not change, so glycine got them to deep sleep sooner without giving them more of it. A glycine maker (Ajinomoto) funded the trial.

Those 3 g are also the dose to use, because that is what the studies tested. A later study found that the same dose eased fatigue and sped up reaction times over three nights of shortened sleep.^[20] One warning comes with it: don't combine glycine with clozapine. How glycine works is known only from rats. There it seems to act on the body-clock region of the brain to widen skin blood vessels and drop core temperature,^[21] the same heat-shedding route the temperature section describes.

Magnesium's evidence is thinner. A review found only three small trials in 151 older adults, and just two of them, with 55 people in total, measured how long it took to fall asleep.^[22] Pooled, magnesium shortened that by about 17 minutes, but with low-certainty evidence.^[22] The review judged the evidence overall too weak for doctors to make well-informed recommendations.^[22]

The two trials behind that number look quite different. The one in healthy older volunteers is the only one with a sleep lab. Its 12 participants aged 60 to 80 got about 6 more minutes of deep sleep than on placebo,^[41] on a dose that built up to about 730 mg of magnesium a day.^{[41, 22]} The other gave 46 older adults with insomnia 500 mg of elemental magnesium a day (as magnesium oxide, 250 mg twice daily) or placebo for 8 weeks.^[23] That is twice the EU upper limit for supplements, a supervised trial dose and not one to copy. Self-rated sleep and the time to fall asleep improved, but without a sleep lab.^[23] The group was also low on magnesium to start, so the result does not stretch to healthy adults who already get enough.

Both of those doses, 500 to 730 mg a day, sit above the EU upper limit of 250 mg a day from supplements.^{[22, 23, 50]} That makes the dose the part to get right. A trial in 155 adults who slept poorly tested exactly 250 mg as bisglycinate, and insomnia scores improved slightly more than on placebo after 4 weeks.^[80] It was a small effect, measured only by questionnaire, with no sleep lab and no data on deep sleep.^[80]

Magnesium bisglycinate is often sold as two actives in one, which holds only partly. One magnesium atom is bound to two glycine molecules, so by weight the compound is about 14 percent magnesium and 86 percent glycine. A dose of 400 mg elemental magnesium therefore carries roughly 2.5 g of glycine, close to the studied 3 g. At 200 mg it is only about 1.2 g, and a label reading "1,000 mg providing 140 mg elemental magnesium" carries about 860 mg. So the two-in-one idea only holds at 400 mg of elemental magnesium, which is above the EU supplement limit.^[50]

Pricier forms do not change that picture, though one shows early signals. Magnesium L-threonate (Magtein) is sold as the superior sleep magnesium, and in the newer of two industry-linked trials, 100 adults taking 2 g a day for 6 weeks got better cognition scores and some better self-rated sleep measures.^[66] Sleep measured by an Oura ring did not differ,^[66] and none of the trials used a sleep lab, so for sleep it is an experiment rather than an upgrade. Magnesium taurate (magnesium bound to taurine, about 9 percent elemental magnesium) has a plausible calming story for both halves; so far its sleep case rests on receptor biology.

Whichever form you pick, two safety points come with any dose. If you have kidney disease, don't take magnesium supplements without your doctor, because weak kidneys can't clear it.^[50] Loose stools are the first sign of too much.^[50] With those two points covered, a sensible way to try both looks like this:

• Magnesium bisglycinate at up to 250 mg of elemental magnesium in the evening, which stays within the EU limit.^[50]
• Glycine at 3 g on its own, about 30 minutes before bed, if you want its effect on falling asleep.^[19]
• No reliance on the glycine inside a normal magnesium dose to reach those 3 g.

Lemon verbena has improved self-rated sleep and perceived stress in human trials, which makes it a reasonable option if you want nights that feel better. Those trials tested lemon verbena alone, measured with questionnaires and a Fitbit rather than brain waves,^{[24, 25]} and no published sleep-lab trial has shown that the combination with zinc deepens sleep, so it is not a proven deep-sleep booster.

Most deep-sleep claims for lemon verbena trace back to one extract, RelaxPLX from the Spanish maker Monteloeder, tested at 400 mg a day. Other lemon verbena extracts are different products, and Robuvit, which sometimes turns up in the same conversation, is made from oak wood.

That 400 mg dose was tested in a trial in which 40 people took RelaxPLX or placebo for 8 weeks.^[24] Sleep-quality scores and perceived stress improved, and blood cortisol fell compared with each group's starting value.^[24] There was no sleep lab, though: the widely quoted "increase in deep sleep" is an estimate from an older Fitbit (Charge 2), based on movement and heart rate.^[24] Even newer Fitbits caught only about half of lab-scored deep sleep in independent testing.^[43] Monteloeder supplied the extract, and three of the six authors work there.^[24]

The later trials look similar. A 90-day trial used questionnaires and a movement tracker and found better self-rated sleep and higher night-time melatonin, and three of its authors work for Monteloeder.^[25] An Iranian trial in insomnia patients found better questionnaire scores with a lemon verbena syrup than with placebo, again without a sleep lab.^[68]

The only sleep-lab data for lemon verbena come from mice. In that study, non-dream sleep rose by roughly 110 to 195 percent over the mice's own baseline across three doses, with no clear dose order.^[28] Dream sleep rose only at the two lower doses.^[28] Those are animal results, so for people the deep-sleep claim still rests on a Fitbit estimate.

Zinc's case follows the same pattern. A review describes how zinc acts on several brain receptors involved in sleep.^[26] In mice, a zinc-yeast extract raised non-dream sleep in a pattern that looked like natural sleep.^[27] In people, the evidence is thin and indirect. One trial in healthy adults found that zinc-rich food helped them fall asleep faster and sleep more efficiently on a wrist tracker, with no sleep lab.^[67]

The products that pair RelaxPLX with zinc bisglycinate, often with other herbs, have not been published as a tested formula. Researcher names that sellers sometimes cite (Buguet, Carrillo-Vico and others) appear in the wider sleep literature, not on a lemon-verbena-plus-zinc deep-sleep trial.

If you try such a product anyway, count all your zinc. The EU upper limit for zinc is 25 mg a day in total, from food and supplements combined.^[50] Long-term higher intakes disturb copper balance and have been linked to anaemia and low white-cell counts, so keep your total at or below that.^[50]

Lemon verbena has its own limits. The EU herbal monograph covers the leaf as a traditional herbal tea, 1 to 2 g of leaf before bed to aid sleep.^[51] Concentrated extracts like the 400 mg trial product fall outside it.^[51] It advises against use in pregnancy, breastfeeding and children, and it rules the herb out if you are allergic to plants of the verbena family.^[51] It also warns that it may impair driving and says to see a doctor if symptoms last more than 2 weeks.^[51]

Taken together, lemon verbena plus zinc is a modestly studied option for self-rated sleep, about level with magnesium glycinate or saffron. Its safety data cover the tea rather than the extracts, its human trials are industry-linked, and none has confirmed deeper sleep in a sleep lab. If falling asleep is your goal, it is a reasonable option to try; for "deep sleep enhancement" specifically, the evidence does not justify paying a premium.

L-theanine, saffron, taurine and melatonin turn up in most consumer sleep formulas, and where they work, they help with falling asleep, anxiety, how you rate your night or the timing of your body clock. None has sleep-lab evidence of deepening sleep in humans.

L-theanine, an amino acid from tea, calms without sedating. At 50 mg, it raised alpha activity (the brain waves of calm wakefulness, not of sleep) compared with placebo in young adults.^[69] In a trial of 30 stressed but healthy adults, 200 mg a day for 4 weeks improved sleep-quality scores.^[29] Three sub-scores (time to fall asleep, disturbance, use of sleep medication) improved more than on placebo.^[29] All of it was self-rated, with no sleep lab.

The other L-theanine studies point the same way. In boys with ADHD, 400 mg a day improved sleep efficiency on a wrist tracker, though the time to fall asleep did not change.^[70] A review concluded that 200 to 400 mg a day may reduce stress and anxiety in people under stress.^[71] That makes L-theanine a good fit for people kept awake by racing thoughts, rather than a deep-sleep supplement.

Saffron extract (from the saffron crocus) also improves self-rated sleep, though over weeks rather than on the first night. Why it helps sleep is not known; better mood is one idea, not a shown mechanism. Two trials with different doses make the case, both mainly with questionnaires. In one, adults who slept poorly took 28 mg a day of the extract affron for 4 weeks and improved their insomnia scores more than the placebo group.^[30] In the other, adults with mild-to-moderate sleep problems and anxiety took 15.5 mg a day for 6 weeks and also scored better, with more time in bed on a wrist tracker.^[65] Both had industry money behind them: the maker of affron paid for the first, and the extract's supplier funded the second study centre.^{[30, 65]}

Neither trial looked at brain waves, which is what a pilot in 52 older adults added. With an EEG headband, which records brain waves at home, 30 mg a day sped up sleep onset.^[64] It did not show more deep sleep in its main results, and its co-authors include staff of the extract maker.^[64] Since the trials ran for 4 to 6 weeks, a fair test means taking saffron daily for at least that long.

So far, taurine's sleep case is theoretical. It acts on calming receptors in lab work, but no human trial of taurine on its own has shown a sleep benefit. Most "taurine and sleep" data come from energy-drink studies that pair it with caffeine. If a formula mixes taurine with glycine and magnesium, any benefit cannot be credited to the taurine; its stronger story is in your cells' power plants (see the mitochondria guide).

Melatonin is a body-clock signal, not a depth booster: it tells your brain it is dark and shifts your timing. A meta-analysis of 19 studies found that it shortened the time to fall asleep by about 7 minutes and added about 8 minutes of total sleep.^[31] Across trials, higher doses and longer use shortened sleep onset a little more but did not improve sleep quality.^[31] The review did not test high against low doses head to head, though.^[31] No sleep-lab evidence shows melatonin adding deep sleep.

That gives melatonin a clear, specific job: jet lag (especially flying east), a delayed sleep schedule, shift work and body-clock disorders. The right time to take it depends on the reason, so plan shift work and body-clock disorders with a doctor, and don't drive or do safety-critical work while drowsy. A low dose of about 0.3 to 1 mg is a reasonable start, not a tested result.

Prescription melatonin is a separate product. In the EU, Circadin (2 mg prolonged-release melatonin) is a prescription medicine for short-term treatment of primary insomnia in people 55 and over, for up to 13 weeks.^[48] It is not the same as a melatonin supplement. The European insomnia guideline allows prolonged-release melatonin for up to 3 months from age 55 and advises against fast-release melatonin for insomnia.^[45]

There are no good long-term safety data for nightly melatonin. A 2025 conference report linked a year or more of use to heart failure; it is observational and unconfirmed, and a later national survey analysis found no such link.^[74] In Germany, the BfR (the federal institute for risk assessment) has warned about health risks from higher-dose melatonin supplements (as of October 2026). The strengths of these four lie in falling asleep, calmer evenings and body-clock timing; for deeper sleep itself, removing what disturbs it is the better bet.

The things that do the most damage to sleep are alcohol close to bedtime, some sleeping pills, drugstore antihistamines, late caffeine and untreated apnea. Removing them lays the foundation, before you add anything.

Alcohol is the most misunderstood of these, because it can add deep sleep. A review of sleep-lab studies found that alcohol shortens the time to fall asleep.^[32] Across doses, ages and sexes, it also tends to add deep sleep in the first half of the night.^[32] As blood alcohol falls, though, the second half breaks up with more waking, and total dream sleep (REM) for the night goes down.^[32] Alcohol also worsens sleep apnea by relaxing the muscles that hold the airway open.

That broken second half is what you feel the next day, and a wearable may score the whole night as poor even when early deep sleep went up. If you drink, reasonable limits are to finish at least 3 hours before bed and keep it to about one drink on a sleep night. With diagnosed apnea, skipping evening alcohol makes sense.

Sleeping pills differ in what they do to deep sleep. Benzodiazepines tend to reduce it, while results for Z-drugs such as zolpidem and zopiclone vary; zolpidem even increased deep sleep in one sleep-lab study in young adults.^[53] Whatever they do to deep sleep, both classes bring tolerance, dependence, falls in older adults and next-day impairment.

The most serious risk is what people do while not fully awake. Since 2019, zolpidem, zaleplon and eszopiclone carry a US boxed warning for complex sleep behaviors such as sleepwalking and sleep-driving, which can cause serious injury or death.^[62] An observational study also found that people prescribed hypnotics died at roughly 3.6 to 5.3 times the rate of matched non-users, rising with dose.^[37] It cannot prove cause, but the signal has not gone away. If you take a Z-drug or benzodiazepine and want to stop, do it slowly under a doctor's care, and add CBT-I at the same time.

Daridorexant works differently: it blocks orexin, the brain's wake-up signal, instead of boosting a calming receptor. In its two phase 3 trials, the large trials that come before approval, it improved staying asleep and falling asleep at 1 and 3 months.^[33] A pooled analysis of those trials found no change in the brain-wave patterns of deep or dream sleep, and fewer markers of a hyper-alert brain.^[40] That analysis did not compare it with Z-drugs.

Whether you need any of these drugs is a question guidelines answer in steps. CBT-I comes first; if it is not enough, a drug can be offered.^[45] The European guideline lists benzodiazepines, Z-drugs and daridorexant for short-term use, with orexin blockers usable for up to 3 months or longer in some cases.^[45] In the UK, NICE recommends daridorexant only if CBT-I has been tried and not worked, or is unavailable or unsuitable.^[49] Either way, sleeping pills are a decision for you and your doctor, not something to start or stop on your own.

Access to daridorexant differs by country (as of October 2026). In the EU, daridorexant (Quviviq) is prescription-only and must not be used with narcolepsy or with strong CYP3A4-blocking medicines.^[46] In Germany, statutory health insurance has covered it for chronic insomnia without a time limit since 11 November 2023.^[47] The G-BA, the body that decides on public insurance benefits, found its added benefit "not proven" in August 2024.^[47]

What you can buy without a prescription needs care too. High-dose melatonin (5 to 10 mg) is not needed: across trials, higher doses did not improve sleep quality, though the review did not compare high and low doses directly.^[31] The melatonin section covers when a low dose makes sense.

Drugstore antihistamine sleep aids (diphenhydramine, doxylamine) are the bigger concern. They sedate through an anticholinergic effect, which blocks a key brain messenger. The AGS Beers list, a US list of medicines that are often a poor fit for older adults, says to avoid them past age 65.^[63] In a study of older adults, the heaviest long-term users of strong anticholinergic drugs had a higher risk of dementia over about 7 years of follow-up, an observational link.^[34]

Stimulants are simpler to deal with. Late caffeine costs you sleep: 400 mg (about four cups of filter coffee) taken 6 hours before bed cut total sleep by more than an hour in a home study.^[38] Nicotine is a stimulant too, so skip it in the evening.

Nightly cannabis for sleep is a poor bet for a different reason. THC can make falling asleep easier at first, but tolerance to that effect is likely, and stopping after regular use often brings a stretch of poor sleep and vivid dreams.

Late meals matter most for people with apnea. In a study of 296 people with sleep apnea, those who habitually ate late had more breathing pauses and less dream sleep than early eaters.^[59] It was an association, not a test of moving dinner earlier, but eating 3 hours or more before bed is a reasonable habit if you have or suspect apnea.

A realistic 30-day plan starts by removing what breaks sleep, then adds temperature, exercise and finally supplements as the fine-tuning, one week at a time so you can tell what helped.

Days 1-7: Measure your baseline and remove what breaks sleep.

• Keep a 7-day sleep diary: bedtime, wake time, alcohol, caffeine cut-off time and how well you slept on a scale of 1 to 5.
• If you have a wearable, log total sleep time, sleep efficiency and its deep-sleep estimate each night.
• Finish alcohol at least 3 hours before bed, or skip it for the 30 days if you can.
• Stop caffeine after noon if you don't know how fast your body clears it.
• Move the last meal to 3 hours or more before bed.
• Drop over-the-counter antihistamine sleep aids such as diphenhydramine and doxylamine.^[63]
• If you take Z-drugs or benzodiazepines nightly, talk to your doctor about whether and how to taper, as a longer-term project; don't stop on your own.

Days 8-14: Add temperature and light.

• Keep the bedroom from getting hot. Around 17 to 20 degrees C (62 to 68 degrees F) with a duvet is a common comfort range; anything from about 13 to 23 degrees C was fine for sleep stages.^[16]
• Take a warm bath or shower of about 40 degrees C for 10 minutes or more, 1 to 2 hours before lights-out, to fall asleep faster.^[15]
• Keep the bath safe: older adults and people with high blood pressure or heart disease stay below 42 degrees C.^[52] No bath after alcohol or a big meal, not alone if you are frail, and get out slowly.^[52] If you have heart disease, are pregnant or have low blood pressure, ask your doctor first.
• Wear bed socks if your feet run cold.
• Get outdoor light in the morning after waking, cloudy days included; it helps set your body clock.
• Dim the lights for the last few hours before bed, and keep the phone out of reach if you can.

Days 15-21: Add aerobic exercise.

• Do 3 to 4 aerobic sessions a week, 30 to 60 minutes each, at an effort where you can talk but not sing.
• Add 1 to 2 interval sessions a week if you like (for example, 10 rounds of 1 minute hard and 1 minute easy); with heart disease, pregnancy or low blood pressure, ask your doctor first.
• Finish hard sessions at least an hour before bed, since vigorous exercise close to bedtime may delay sleep.^[18]
• The measured deep-sleep effect comes from single evening workouts,^[18] so treat the weekly plan as a sensible training habit rather than a tested dose.
• Morning sessions help your clock too: in one study, five days of morning exercise moved the body clock about 0.6 hours earlier.^[58]

Days 22-28: Fine-tune with supplements once the rest is in place.

• Magnesium bisglycinate at up to 250 mg of elemental magnesium in the evening, the EU upper limit for supplements.^[50] With kidney disease, take it only if your doctor agrees.^[50] At this dose, a 2025 trial found only a small improvement in insomnia scores, with no sleep lab,^[80] while the older trials used 500 to 730 mg.^[22]
• Glycine at 3 g about 30 minutes before bed, the dose from the glycine trial.^[19] It tastes sweet and mixes well into water or tea. Don't combine it with clozapine.
• Optional: L-theanine at 200 mg if racing thoughts at bedtime are your main problem.^[29]
• Melatonin only if you have a body-clock reason such as jet lag, and then at a low dose.^[31]; prescription prolonged-release melatonin for insomnia is a separate decision for your doctor^[48]

Talk to your doctor before starting any of these, especially if you take medication.

Days 29-30: Take stock.

• Compare week 1 with week 4 in your diary and in 7-day wearable averages, never single nights.
• Look for rising sleep efficiency (the share of time in bed you spend asleep) and better self-rated sleep.
• If your wearable shows more deep sleep, read it as a direction, not a precise number.
• If nothing has moved, the cause may be medical, and the signs below say when to see a doctor.

When to see a sleep doctor.

• Loud snoring with witnessed breathing pauses and daytime sleepiness, even with this plan in place, or a score of 3 or more on STOP-BANG, a short sleep-apnea screening questionnaire: see a doctor about apnea.^[56] Questionnaires cannot diagnose it, so the doctor decides between a home test and a lab study.^[56] A negative home test does not rule apnea out if symptoms persist, and people with significant heart or lung disease, a past stroke, long-term opioid use or severe insomnia need a lab study rather than a home test.^[56]
• Insomnia that lasts 3 months or more despite the plan: structured CBT-I with a clinician or a validated digital programme.^[45] In Germany, digital CBT-I apps listed as a DiGA (a prescribable health app) can be prescribed at statutory insurance cost, and the BfArM directory shows which ones are listed.^[82] In the UK, NICE recommends the Sleepio app for insomnia in primary care.^[81] (as of October 2026).
• Acting out your dreams: see a neurologist or sleep specialist soon, because acting out dreams can be an early sign of other brain conditions^[76] Until then, make the bedroom safe: move sharp or breakable objects away from the bed, and a partner may need to sleep separately.
• A restless-legs pattern: get ferritin (your iron stores) and transferrin saturation checked, and don't start iron without a blood test and a doctor. The AASM guideline advises iron (tablets or IV) when ferritin is 75 ng/mL or lower or transferrin saturation is below 20 percent, and IV iron when ferritin is between 75 and 100.^[42] These cut-offs are expert consensus, not tested in trials.^[42] For medicines, the guideline favours gabapentin-type drugs and advises against routine use of older dopamine drugs such as pramipexole.^[42]

If none of those signs apply, let the diary make the call at the end of the month. A reasonable rule is to keep what came with better-rated nights and to drop any supplement that made no visible difference, rather than adding a new one. A short list you have tested on yourself is the mark of a good routine.