Melatonin: Optimal Dosing and the Low-Dose Debate

Key Takeaways

Walk into any pharmacy in the United States and the smallest melatonin tablet on the shelf is usually 3 mg. The common ones are 5 and 10. The gummies aimed at children are 1 to 3 mg each, and children do not stop at one. Now look at the studies that established what melatonin does to human sleep: the dose was 0.3 mg. A thirtieth of a 10 mg tablet.

That gap is not a conspiracy. It is a combination of regulatory history — melatonin is a supplement in the US and a medicine almost everywhere else — and a genuine scientific argument about what the hormone is for. This article lays out the dose-ranging trials, the meta-analyses, the clinical guidelines and the phase-response curves, and then says plainly what dose to take, when, and for which problem. If you want the substance profile first, it is here.

What a Physiological Dose Means

At night, a healthy adult's plasma melatonin sits somewhere between 50 and 150 picograms per millilitre. In the 1994 study that started the low-dose line of research, 20 healthy men took 0.1, 0.3, 1.0 or 10 mg at midday. The two lowest doses produced peak serum concentrations within the normal night-time range; 1 and 10 mg pushed far past it. Yet every dose, including 0.1 mg, shortened the time to fall asleep, lengthened sleep and lowered body temperature compared with placebo.

Two years later the same group gave 12 young adults 0.3 or 1.0 mg two to four hours before their usual bedtime. Both doses shortened sleep latency and the time to reach stage 2 sleep. Sleep architecture was unchanged, 0.3 mg raised serum melatonin to a normal nocturnal 113 pg/ml, and there was no hangover the next morning.

The paper that defined the debate came in 2001. Fifteen insomniacs over 50 and fifteen normal sleepers each took a week of 0.1, 0.3 and 3.0 mg and placebo, under polysomnography. In the insomniacs, 0.3 mg restored sleep efficiency and brought plasma melatonin back into the normal range. The 3 mg dose also improved sleep — but it induced hypothermia and kept plasma melatonin elevated into the daylight hours. In the normal sleepers, nothing happened at any dose. The authors' conclusion was the one the supplement industry never adopted: the physiological dose is the right one, and the pharmacological dose carries a tail.

What Melatonin Does to Sleep — Honestly

Three meta-analyses have pooled the randomised trials, and they agree with each other more than their press coverage suggests.

Seven minutes faster to sleep. Eight more minutes asleep. These are real effects, statistically robust, and small. The 2005 general-medicine review put it bluntly: melatonin is not effective for most primary sleep disorders in the short term, is effective for delayed sleep phase, and appears safe for up to three months.

The 2017 American Academy of Sleep Medicine guideline on insomnia pharmacotherapy reached the same place. Its recommendation: clinicians should not use melatonin to treat sleep-onset or sleep-maintenance insomnia in adults. The strength is "weak", and it sits alongside trazodone, diphenhydramine and valerian in the same category — things people take for sleep that the evidence does not support for insomnia.

That is the context in which the low-dose question has to be read. If you have insomnia with a normal body clock, the debate about 0.3 versus 3 mg is mostly about which dose gives you seven minutes with fewer side effects. For that problem, the real answer lies in cognitive behavioural therapy for insomnia, and in protocols aimed at sleep depth such as the Deep Sleep Stack — which, as it happens, is fully compatible with a low melatonin dose.

Where Melatonin Is Genuinely Strong

Now change the problem. Suppose your melatonin onset is at 1 a.m. and your natural sleep time is 3 a.m., and the world expects you up at 7. That is delayed sleep-wake phase disorder, and here melatonin stops being a weak hypnotic and becomes a clock-setting drug.

A meta-analysis of nine placebo-controlled trials in delayed sleep phase — five in adults, four in children — found that melatonin advanced the dim-light melatonin onset by 1.18 hours, moved sleep-onset clock time 0.67 hours earlier and reduced sleep latency by 23 minutes. The largest modern trial randomised 116 adults with a confirmed delayed melatonin onset to 0.5 mg fast-release melatonin or placebo one hour before desired bedtime, plus behavioural scheduling, for four weeks. Sleep began 34 minutes earlier on melatonin than on placebo, and clinicians rated 53% improved versus 24%. The 2015 AASM circadian guideline endorses strategically timed melatonin for exactly this disorder, as well as for blind adults with non-24-hour rhythms.

Jet lag is the other robust indication. The Cochrane review found that melatonin taken at destination bedtime reduced jet lag in nine of ten trials after flights crossing five or more time zones. The number needed to treat was two. Doses from 0.5 to 5 mg all worked.

The pattern is consistent: melatonin is a chronobiotic first and a hypnotic second. Its effect size is large when the problem is timing and small when the problem is sleep itself.

The Phase-Response Curve: Why Timing Beats Dose

Melatonin moves the clock along a curve that is roughly the mirror image of light's. Taken in the late afternoon or early evening, before your own melatonin has started to rise, it advances the clock — you will get sleepy earlier and wake earlier. Taken in the morning, it delays it. This was shown in the early 1990s with physiological doses and is the basis for every circadian use of the hormone.

The dose question was tested directly in 2010. Thirty-four healthy adults took 0.5 mg melatonin for three days at one of several clock times, and the resulting phase-response curve was compared with one previously built using 3 mg on the same protocol. The maximum advances with 0.5 mg came when it was taken in the afternoon, two to four hours before the dim-light melatonin onset — about nine to eleven hours before the midpoint of sleep. Crucially, the optimal time for 0.5 mg was later in the day than for 3 mg, and when each dose was taken at its own best time, the advances were of similar size.

Read that again, because it is the most practically useful finding in the literature. A 0.5 mg dose, timed to the afternoon, moves the clock about as far as 3 mg. The higher dose does not buy you more shift; it buys you a longer sedative tail, and because its optimum is earlier in the day it is more likely to make you drowsy at 4 p.m.

One honest caveat from the 116-person trial: in the subset whose melatonin onset was re-measured afterwards, it had not shifted more on melatonin than on placebo. The 0.5 mg dose taken one hour before bed was helping people fall asleep more than it was moving their clock. That is consistent with the phase-response curve — one hour before bed is late for a phase advance. To move the clock, take it earlier.

The Case for Higher Doses

The low-dose argument is not the whole story, and it would be dishonest to pretend otherwise.

The 2013 meta-regression across 19 trials found that higher doses and longer trial durations were associated with larger reductions in sleep latency and larger gains in total sleep time. In the Cochrane jet-lag review, 5 mg produced faster sleep onset and better sleep than 0.5 mg, though doses above 5 mg added nothing and a 2 mg slow-release product was relatively ineffective. For the specific job of knocking yourself out at a strange bedtime in a strange time zone, more melatonin is a stronger sedative.

Then there is prolonged-release 2 mg, the form licensed as a medicine in Europe for insomnia in adults over 55. In a primary-care trial of 354 people aged 55 to 80, three weeks of it produced a responder rate of 26% versus 15% on placebo, with sleep latency falling 24 minutes versus 13. A 791-person trial that ran for six months found the benefit was maintained with no tolerance and no clinically relevant safety signal. This is not a physiological dose — 2 mg is well above nocturnal levels — but the slow-release profile keeps the curve low and flat through the night, which is a different strategy from a 10 mg bolus.

So the fair summary is a trade-off:

• 0.3–0.5 mg: reproduces physiology, shifts the clock as effectively as higher doses when timed correctly, no morning hangover, no hypothermia. The dose for circadian problems.
• 2 mg prolonged-release: the evidence-based choice for poor sleep maintenance in older adults, with six-month safety data.
• 3–5 mg immediate-release: a stronger, longer hypnotic. Reasonable for a few nights of jet lag. For nightly use, it is the dose most likely to leave melatonin in your blood at breakfast.
• 10 mg and above: no efficacy advantage in any trial, and a 40% higher rate of drowsiness, headache and dizziness in a pooled analysis of high-dose studies.

What Is Actually in the Bottle

A dose argument assumes the dose on the label is the dose in the tablet. In 2017 a Canadian laboratory analysed 31 commercial melatonin products by liquid chromatography. Measured content ranged from 83% below to 478% above the labelled amount. Lot-to-lot variation within a single product reached 465%. More than 71% of products were not within 10% of their label. Eight of the 31 — a quarter — contained serotonin, a controlled substance in some jurisdictions and a compound nobody intended to buy.

The practical consequence is that a "0.5 mg" tablet from an untested brand might deliver 0.1 mg or 2.9 mg, and you have no way of knowing which. If you are using melatonin for its clock-shifting effect, where precision of dose and timing is the whole point, buy a pharmaceutical-grade product or one with a published certificate of analysis.

Safety, Including the Part About Children

Melatonin is, by the standards of sleep drugs, benign. A systematic review of 37 placebo-controlled trials at doses from 0.15 to 12 mg found the commonest adverse events were daytime sleepiness (1.7%), headache, other sleep disturbance, dizziness and hypothermia (each under 1%), with no life-threatening events. The high-dose meta-analysis found no excess of serious adverse events but more drowsiness, headache and dizziness. What both reviews stress is that long-term randomised data are scarce: most trials lasted four weeks, few passed six months. "Safe for up to three months" is the defensible claim.

Children are a different conversation. Between 2012 and 2021, US poison centres received 260,435 reports of paediatric melatonin ingestion. The annual number rose 530%. Melatonin went from 0.6% of all paediatric ingestions to 4.9%. Most were unintentional ingestions by children under five — gummies look like sweets — and the rise in hospitalisations and serious outcomes followed the same group. Five children needed mechanical ventilation. Two died. The AASM circadian guideline supports timed melatonin in children only for specific circadian disorders, under supervision, and recommends against it in older adults with dementia. If there is melatonin in your house, treat it like any other medicine.

Drug interactions are few but real: case reports with warfarin and antiepileptic drugs, additive sedation with alcohol and benzodiazepines, and a large increase in melatonin exposure with fluvoxamine, which blocks the enzyme that clears it.

How to Dose, by Problem

You fall asleep too late and wake too late. This is a circadian problem and melatonin's best use. Take 0.5 mg in the afternoon or early evening, two to four hours before your own melatonin onset — for most late sleepers, four to six hours before the time they actually fall asleep. Combine it with the light and schedule anchors in the Circadian Optimization Protocol, which do more of the work than the tablet. Expect to shift by 15–30 minutes a week.

You want to fall asleep sooner tonight. 0.5–1 mg, 30 to 60 minutes before the bedtime you want. Expect seven minutes, not an hour. If this is a nightly problem, you have insomnia, and the fix is elsewhere.

You are over 55 and wake through the night. Prolonged-release 2 mg, one to two hours before bed, is the form with the trial evidence.

You are crossing five or more time zones eastward. 0.5–3 mg at destination bedtime for the first few nights. Do not take it earlier in the day, which causes sleepiness and delays adaptation.

You have insomnia with a normal body clock. Melatonin is not your drug. The guideline that says so is the one written by the people who treat insomnia for a living.

Frequently Asked Questions

Does melatonin stop working if I take it every night? The six-month trial of prolonged-release 2 mg found no tolerance. There is no evidence that exogenous melatonin suppresses your own production. What the data do not cover is years of nightly use.

Is 10 mg dangerous? Not acutely, in adults. It is simply no more effective than 3 mg and more likely to leave you groggy. The pooled high-dose analysis found a 40% higher rate of drowsiness, headache and dizziness.

Should I take it with food? No specific requirement. Timing relative to your clock matters far more than timing relative to meals.

What about melatonin for children with autism or ADHD? That is a clinical question with its own literature and should be handled by a paediatric sleep specialist. This article covers healthy adults.

Can I combine it with magnesium or glycine? Yes. Magnesium L-threonate and glycine act on sleep depth and body temperature through separate mechanisms, and low-dose melatonin fits alongside either.

Where do I start? With your clock, not the tablet. The Circadian Optimization Protocol puts light, wake time and meal timing first and melatonin where the evidence says it belongs.

Related Research

• Melatonin
• Circadian Light: Managing Blue Light and Morning Sun
• The Circadian Optimization Protocol: Light + Melatonin + Timing
• The Circadian Optimization Protocol
• The Deep Sleep Stack: Magnesium Threonate + Apigenin + L-Theanine + Glycine