Sleep supplements: what the evidence shows

Most sleep supplements rest on small, short trials with modest results. Melatonin has the most evidence, and it is better at moving the body clock than at sedating: across 19 trials pooled in 2013 it brought sleep on about 7 minutes sooner. The rest range from thin to inconsistent, labels are often inaccurate, and for insomnia that has lasted months the recommended first treatment is CBT-I, not a supplement.

11 min readPublished 14 October 2026By SheepsHow we write these pages

Do sleep supplements work?

Some do a little, none does much, and for most of them nobody yet knows. The trials behind the bottles on a pharmacy shelf are mostly small, short and of mixed quality, and when they are pooled, the differences against placebo are measured in minutes, not hours.

Melatonin is the best studied. Pooled across 19 placebo-controlled trials in people with sleep disorders, it brought sleep on 7 minutes sooner and added 8 minutes of sleep on average — a real effect, which the authors call modest (Ferracioli-Oda and colleagues, 2013). Its stronger case is timing — jet lag, and a body clock that runs late — and the melatonin page goes through both.

For the others, the trials point in different directions, several of the most popular rest on a handful of studies, and some of those studies were paid for by companies with a stake in the result. The table takes them one at a time.

What does each supplement do, according to the trials?

What each one is, what was tested and in whom, what was found, and what to watch for. A dose is given only where a study used it; it is not a recommendation.

Ten sleep supplements against their evidence. Checked against the reviews, trials and health agencies named, October 2026.
What it is	What was tested, in whom	What was found	Cautions
Melatonin — the hormone the brain releases in darkness	19 placebo-controlled trials pooled in 2013: 1,683 adults and children with insomnia, a late body clock or another sleep disorder, taking 0.1 to 5 mg	Sleep came 7 minutes sooner and lasted 8 minutes longer than on placebo. Better evidence for jet lag (Cochrane review, 2002) and for a late body clock (AASM, 2015): more on melatonin	Prescription-only in the UK (NHS), a dietary supplement in the US (NCCIH). Not usually recommended in pregnancy; interacts with warfarin and other medicines (NHS)
Magnesium — a mineral, sold as oxide, citrate, glycinate and other forms	12 randomised trials in adults, reviewed in 2026, using different forms, doses, lengths and measures	Inconsistent: possibly modest self-rated benefits for some adults, with certainty low to very low; the review does not support it as a routine insomnia treatment (more on magnesium)	Diarrhoea at high doses. Risk of a dangerous build-up with kidney disease; interferes with some antibiotics and osteoporosis drugs (NIH Office of Dietary Supplements)
Valerian — an extract of valerian root	Herbal medicines for insomnia pooled in 2015: 14 trials, 1,602 people; 18 valerian trials pooled in 2010	No significant difference from placebo on any of 13 measures (2015). People more often said they slept better, but the time they took to fall asleep did not change (2010)	More side effects than placebo in the 2015 review; long-term safety uncertain (NCCIH)
Chamomile — a daisy-family flower, taken as tea or extract	Included in the 2015 review of herbal medicines in people with insomnia	No significant difference from placebo; NCCIH finds no conclusive evidence from trials either way	Allergic reactions, especially in people allergic to ragweed or related plants (NCCIH)
Ashwagandha — an extract of the plant Withania somnifera	5 placebo-controlled trials, 400 adults, pooled in 2021	A small improvement in overall sleep, larger in people with insomnia, in trials using 600 mg a day or more, and over 8 weeks or longer	Avoid in pregnancy and breastfeeding; not advised with thyroid or autoimmune disorders or before surgery; rare liver injury; several drug interactions (NCCIH)
Glycine — an amino acid	3 trials in all, a 2026 review found; a 2012 study gave 3 g at bedtime to 10 healthy men whose sleep was cut short for 3 nights	Each trial reported a benefit on at least one measure; in the 2012 study, less daytime fatigue	Too few trials to judge. The 2012 study was run by a company’s researchers on its own employees
L-theanine — an amino acid found in tea	19 randomised-trial reports, 897 people, pooled in 2025; few tested L-theanine on its own	Small improvements in self-rated time to fall asleep, sleep quality and daytime functioning	Short trials; the right dose and length are still unsettled (2025 review)
Tart cherry — Montmorency cherry juice or concentrate	7 intervention studies, reviewed in 2025, with different doses, lengths and people	3 of the 7 reported better sleep on some measure; the reviewers call the evidence limited (more on the cherry drink)	It is fruit juice, sugar and all; the drink page gives the amounts
Lavender oil capsules — a standardised oral preparation, not the oil for a diffuser	One proprietary capsule (Silexan), 80 mg a day for 10 weeks, in 212 people with anxiety; analysed in 2019	Sleep improved more than on placebo — almost entirely, the analysis found, because anxiety did	Diarrhoea, headache, nausea or burping; may add to sedatives (NCCIH). The analysis was funded by the pharmaceutical company Dr Willmar Schwabe
CBD (cannabidiol) — a compound from cannabis that does not cause a high	12 clinical studies of cannabinoids for sleep disorders, reviewed in 2020	Insufficient evidence to support routine use for any sleep disorder	Associated with liver injury, male reproductive harm and interactions with other drugs (NCCIH)

If a busy mind is what keeps you up, Sheeps gives it a story with nothing to follow.

For chronic insomnia in adults, the American Academy of Sleep Medicine’s 2017 guideline on sleep medicines suggests that clinicians not use melatonin, valerian, tryptophan or the antihistamine diphenhydramine, whether the trouble is falling asleep or staying asleep (Sateia and colleagues, 2017).

Each of those is a weak recommendation, and the guideline is careful about what that means: it reflects low certainty, and “should not be construed as an indication of ineffectiveness”. The evidence was too thin or too mixed to recommend them — which is not the same as evidence that they do nothing.

What the American College of Physicians (Qaseem and colleagues, 2016) and the American Academy of Sleep Medicine (Edinger and colleagues, 2021) recommend first, for insomnia that has lasted months, is cognitive behavioural therapy for insomnia (CBT-I).

Is there a best supplement for sleep?

No, because there is no single sleep problem. What evidence there is sorts the supplements by the problem they might fit:

A body clock in the wrong place. Melatonin’s best case. A Cochrane review found that melatonin taken close to bedtime at the destination reduced jet lag after flights across 5 or more time zones (Herxheimer and Petrie, 2002), and the AASM suggests timed melatonin for delayed sleep–wake phase disorder, a clock that runs hours late (Auger and colleagues, 2015). Does melatonin work for sleep? has the detail, including why timing matters more than dose.
Anxiety that keeps you awake. In a reanalysis of one lavender-capsule trial, 212 people with anxiety, sleep got better mostly because anxiety did (Seifritz and colleagues, 2019). If worry is what keeps you up, the worry is the thing to work on: racing thoughts at night.
Ordinary trouble falling asleep. Nothing on the shelf has strong evidence, and magnesium’s is low in certainty (magnesium for sleep). The techniques in how to fall asleep cost nothing to try.
Insomnia that has lasted months. CBT-I, not a supplement — see the last section.

And before any of them, look at the two chemicals most evenings already contain. Caffeine is still in the body at bedtime, hours after the last cup (caffeine has a very long tail), and a nightcap sedates rather than brings ordinary sleep (alcohol is a sedative, not a sleep aid).

Why can’t you take a supplement label at its word?

Because, in the US at least, no regulator checks it before the product goes on sale. The Food and Drug Administration has no authority to approve dietary supplements before they are marketed; making sure a product is safe, and that its claims are truthful, is first the manufacturer’s job (FDA, 2026).

When researchers have measured what is in the bottle, label and contents have often parted company:

In 31 melatonin supplements tested in 2017, the melatonin ranged from about a sixth of the stated amount to nearly six times it, batches of the same product differed widely, and 8 contained serotonin, which the authors count as a contaminant (Erland and Saxena, 2017).
In a 2023 study of 25 melatonin gummy products sold in the US, only 3 held within a tenth of the amount on the label; a serving contained anywhere from 1.3 to 13.1 mg, and 5 of the products also contained CBD (Cohen and colleagues, 2023).

In the UK, melatonin is a prescription-only medicine; the NHS says supplements sold in other countries are not authorised for sale there, and advises against ordering melatonin online (NHS, 2023). None of this means a particular bottle is wrong. It means the dose in a trial and the dose in your hand may not be the same thing.

How do you read a supplement study?

Five questions sort most of them.

How many people, and for how long? All the evidence on glycine and sleep comes to 3 trials (Fan and colleagues, 2026), and a 2012 study of it followed 10 men over 3 nights (Bannai and colleagues, 2012). Studies that size can only find large, quick effects, and they will miss rare harms.
Who was tested? A result in good sleepers says little about insomnia, and the reverse. The 10 glycine volunteers had no sleep problems until the researchers cut their sleep by a quarter.
What was measured? How people rate their sleep and what a recording shows can disagree. In 18 valerian trials, people more often said they slept better, yet the time they took to fall asleep differed from placebo by 0.7 minutes on average (Fernández-San-Martín and colleagues, 2010). Across 12 magnesium trials, results disagreed between questionnaires, sleep EEG, actigraphy and consumer wearables (Lopresti and colleagues, 2026).
Compared with what? People with insomnia improve on placebo too. Across 32 drug trials for insomnia with 3,969 participants, the placebo groups achieved roughly two-thirds of the improvement seen with the drugs, on sleep-laboratory recordings as well as in questionnaires (Winkler and Rief, 2015). A study with no placebo group cannot tell the supplement from the expectation.
Who paid? Industry money does not make a result wrong, but it is worth knowing. The 2019 lavender analysis was funded by the pharmaceutical company Dr Willmar Schwabe (Seifritz and colleagues, 2019), and the 2012 glycine study was run by researchers at the Japanese company Ajinomoto, on volunteers who worked there.

Who should be careful with sleep supplements?

Anyone who takes other medicines, and anyone in the groups below, should talk to a doctor or pharmacist before starting one. A plant or a mineral can still interact with a prescription.

Pregnancy and breastfeeding. The NHS does not usually recommend melatonin in pregnancy, because little is known about its effect on babies (NHS, 2023). NCCIH advises avoiding ashwagandha in pregnancy and while breastfeeding, and says little is known about lavender in either.
Children. The American Academy of Sleep Medicine advises parents to talk to a health professional before giving a child melatonin or any supplement, and to keep melatonin out of reach like any other medicine (2022).
Other medicines. The NHS lists warfarin, some antidepressants, the epilepsy drug carbamazepine and blood-pressure medicines among those that interact with melatonin. Magnesium interferes with some antibiotics and osteoporosis drugs (NIH Office of Dietary Supplements); ashwagandha may interact with thyroid, diabetes, blood-pressure, sedative and immune-suppressing medicines (NCCIH); and CBD has been associated with interactions with other drugs (NCCIH).
Kidney disease. The kidneys clear excess magnesium; when they work less well, the risk of magnesium toxicity rises (NIH Office of Dietary Supplements).
Liver, thyroid and autoimmune conditions, and surgery. The NHS asks people with liver or kidney problems or an autoimmune condition to tell their doctor before taking melatonin; NCCIH advises against ashwagandha for people with autoimmune or thyroid disorders, or who are about to have surgery.
Allergies. Chamomile can cause allergic reactions, especially in people allergic to ragweed or related plants (NCCIH).
Two to leave alone without advice. Kava, a herb sometimes taken for sleep, has been linked to liver injury that is sometimes serious or fatal, and L-tryptophan may cause life-threatening serotonin toxicity alongside some medicines (NCCIH, 2024).

What if it’s insomnia?

If trouble sleeping has happened on most nights for months and is affecting your days, take it to a doctor rather than to a shelf. For chronic insomnia, the first-line treatment in the guidelines of the American College of Physicians (Qaseem and colleagues, 2016) and the American Academy of Sleep Medicine (Edinger and colleagues, 2021) is cognitive behavioural therapy for insomnia, CBT-I — not a sleeping pill, and not a supplement.

For an ordinary restless evening, start with what happens around bed: a bedtime routine for adults and how to fall asleep go through what has evidence behind it. The supplements on this page are, at best, small additions to that, and none of them is a substitute for care.