Here is a sentence that sounds like a contradiction and is not.
The American Academy of Sleep Medicine, the professional body that writes US sleep medicine’s clinical practice guidelines, advises clinicians against melatonin for insomnia. The same organization endorses melatonin for a different set of sleep problems. Both positions are current and published. Neither is a mistake.
The difference is what melatonin is. The European Medicines Agency describes it on its own assessment pages as a naturally occurring hormone involved in coordinating the body’s sleep cycle, with blood levels that normally rise after the onset of darkness. It is a timing signal. It tells your physiology what hour it thinks it is.
The sleep aisle sells it as sedation. Those two guidelines look contradictory only if you accept that framing, and the whole subject gets clearer the moment you drop it.
Two guidelines, one molecule
In 2017 the AASM published its Clinical Practice Guideline for the Pharmacologic Treatment of Chronic Insomnia in Adults in the Journal of Clinical Sleep Medicine. On the academy’s own three-page summary of that guideline, melatonin appears exactly once, in the section headed Not Recommended. The recommendation statement begins “We suggest that clinicians not use the following drugs for the treatment of sleep onset or sleep maintenance insomnia (versus no treatment) in adults” and then names six: diphenhydramine, melatonin, tiagabine, trazodone, L-tryptophan and valerian. The recommendation carries a weak strength rating.
The agents that guideline does suggest are all prescription drugs, and the evidence behind them is modest too: high quality for triazolam, moderate for temazepam, low for zaleplon and doxepin, very low for four more. Every one of those recommendations is also weak. This is not a field with a confident answer.
Two years earlier, in 2015, the same organization published its Clinical Practice Guideline for the Treatment of Intrinsic Circadian Rhythm Sleep-Wake Disorders. That document reaches the opposite conclusion about the same molecule. Its authors write that the available data allowed positive endorsement, at a second-tier degree of confidence, of strategically timed melatonin for delayed sleep-wake phase disorder, for blind adults with non-24-hour sleep-wake rhythm disorder, and for children and adolescents with irregular sleep-wake rhythm disorder alongside comorbid neurological disorders.
Note the adverb. Strategically timed. The endorsement is not of melatonin, but of melatonin given at a particular hour relative to a person’s own body clock.
The same guideline also recommends against melatonin for one population, elderly patients with dementia, at a second-tier degree of confidence. It is an explicit negative inside a supportive document, and it rarely survives into consumer coverage.
So the reconciliation is simple. For a person who cannot fall asleep or stay asleep, US sleep medicine says the evidence does not support melatonin. For a person whose body clock runs on a different schedule than their life demands, it says timed melatonin has support.
Seven minutes, and the dose question
The size of melatonin’s effect on ordinary sleep surprises most people.
A 2013 meta-analysis in PLoS One by Eduardo Ferracioli-Oda and colleagues pooled 19 randomized placebo-controlled trials covering 1,683 adults and children with primary sleep disorders. Melatonin cut the time to fall asleep by a weighted mean of 7.06 minutes, 95 percent confidence interval 4.37 to 9.75. It added 8.25 minutes of total sleep, interval 1.74 to 14.75. Sleep quality improved by a standardized mean difference of 0.22, a small effect by convention. The analysis had NIH extramural support.
Those results were statistically significant. Seven minutes is both a genuine result and a very modest one. The authors said as much, writing that melatonin’s effects are modest, that they do not appear to fade with continued use, and that the absolute benefit against placebo is smaller than that of other insomnia drugs.
A 2023 network meta-analysis in Sleep Medicine Reviews covering 69 studies, 17,319 patients and 20 drugs ranked orexin receptor antagonists first on every sleep measure it compared, with melatonin receptor agonists behind them. That class means prescription drugs such as ramelteon, not supplement-aisle melatonin. Related mechanism, different evidence.
Timing beats dose, and two research lines point different ways
If melatonin works as a clock signal, the hour should matter more than the milligrams. One of the cleanest tests of that was published in 2010 in the Journal of Clinical Endocrinology and Metabolism by Helen Burgess and colleagues, then at Rush University Medical Center.
Thirty-four healthy adults aged 18 to 42 each spent two five-day laboratory sessions on a double-blind, counterbalanced protocol. The researchers built a phase response curve, a map of how far the body clock shifts depending on the hour a dose is given, for 0.5 mg and compared it with their earlier curve for 3.0 mg.
The largest clock advances occurred when 0.5 mg was taken in the afternoon, two to four hours before the dim light melatonin onset, or nine to eleven hours before the midpoint of sleep. And the optimal administration time is later for the lower dose, so that when each dose is given at its own best hour, both produce shifts of similar size.
Read that twice. Six times the dose did not produce six times the shift. It produced a similar shift at a different hour.
There is a complication, and it should be reported rather than tidied away. The 2013 meta-analysis found the opposite pattern on a different outcome: higher doses and longer trials showed greater effects on sleep latency and total sleep time, while dose and duration made no difference to sleep quality. The two bodies of work point in opposite directions because they measure different things. Burgess measured clock shifting. Ferracioli-Oda measured sedation and sleep duration. Both can be true, and neither resolves the other.
One detail cuts through much of this. The dim light melatonin onset is a laboratory measurement, taken from timed saliva or blood samples under dim light. Nobody can guess their own from a bedside clock, so the hour that worked best in Burgess’s experiment is anchored to a number most people have never had measured.
The tiers, plainly. A small but statistically significant effect on how fast people fall asleep: established. A clinically meaningful effect on chronic insomnia: not established, and specifically not recommended by the leading US professional body. An effect on circadian rhythm disorders when the timing is right: endorsed at a second-tier level of confidence.
A supplement here, a prescription medicine in the European Union
The same molecule sits in two different regulatory categories on two sides of the Atlantic.
In the United States, melatonin is sold as a dietary supplement: no approved drug application behind the bottles, no federal dose ceiling, no prescription requirement at any strength. The AASM’s own health advisory says plainly that melatonin is not under FDA oversight the way over-the-counter and prescription medications are.
Australia treats melatonin as a prescription-only medicine, with exceptions in specific circumstances for adults. In the European Union, a 2 mg prolonged-release melatonin tablet called Circadin has been a centrally authorized medicine since the European Commission granted marketing authorization on 29 June 2007. Its authorized indication is narrow: short-term treatment of primary insomnia characterized by poor sleep quality, in patients aged 55 or over. The European Medicines Agency states that the medicine can only be obtained with a prescription.
The approval data show what a regulator accepted. Across three main trials in 681 patients over 55, after three weeks 32 percent of those on the medicine reported significant improvement in sleep quality and daytime functioning, against 19 percent on placebo. The EMA’s human medicines committee noted the effect was small but judged the benefits to outweigh the risks.
A 13-point difference over placebo, in a selected population aged 55 and over, for a 2 mg prolonged-release formulation, is the evidence that cleared a European prescription bar. It is not evidence about a 10 mg gummy taken by a 34-year-old.
What is in the bottle
Every dose discussion above assumes the label is accurate. That assumption has been tested repeatedly and has not held up well.
A 2017 analysis in the Journal of Clinical Sleep Medicine by Lauren Erland and Praveen Saxena at the University of Guelph measured 31 commercial melatonin supplements by liquid chromatography with mass spectrometry. Against the label, measured content fell as far as 83 percent short and ran as far as 478 percent over. Variation between lots of the same product reached 465 percent. More than 71 percent of the supplements fell outside a 10 percent margin of their label claim. In eight of them the researchers found serotonin, a related compound used as a controlled substance in treating several neurological disorders, at 1 to 75 micrograms. This is the study the AASM advisory points to when it tells parents melatonin has ranged from less than half to more than four times the labeled amount.
A second example comes from Australia, and its context has to be stated carefully because it is usually dropped. The Therapeutic Goods Administration published a safety advisory, updated 29 January 2026, after its laboratories tested eleven imported melatonin products. Measured content ran from 37 percent of label at the low end to 417 percent at the high end, and one product contained no detectable melatonin. The regulator reported significant discrepancies between actual and labelled content, warned about accidental overdose especially in children, and flagged increased seizure frequency as a concern for children with seizure disorders.
Now the qualifier. The TGA headlines the advisory as a warning about imported counterfeit melatonin products, and its body calls them unregistered therapeutic goods, meaning no Australian registration, often bought online or from local retailers. Melatonin is a prescription-only medicine in Australia, except in specific circumstances for adults. Those assays therefore measured products moving outside a regulated channel. They are not a random sample of lawfully marketed supplements, and anyone citing the 417 percent figure without that sentence is overstating it.
Third-party verification exists but is thin. The AASM advisory, adopted in September 2022, reported that only four melatonin products, all at 3 mg or 5 mg, carried the USP Verified Mark. Counts like that change, so treat it as a 2022 figure. Article 325 covers gummies as a formulation.
Children, and the figure that reframed the risk
The pediatric numbers are where this stops being academic.
In March 2024 the CDC published a report in Morbidity and Mortality Weekly Report by Devin Freeman and colleagues on emergency department visits for unsupervised pediatric melatonin ingestion. Drawing on 295 surveillance cases in the National Electronic Injury Surveillance System Cooperative Adverse Drug Event Surveillance project, the authors estimated 10,930 ED visits among children aged 5 years or younger between 2019 and 2022, with a 95 percent confidence interval of 7,609 to 14,251.
That was 7.1 percent of all emergency department visits for unsupervised medication exposures in that age group. Children aged 3 to 5 accounted for 52.4 percent of visits and children aged 0 to 2 for the rest. Gummies were the most commonly documented dosage form at 47.3 percent, and flavored products featured in at least half. Where the container was recorded, roughly three quarters came from a bottle.
Two numbers from that report belong side by side. Most of these children were fine: 93.5 percent of visits did not end in hospitalization, confidence interval 89.3 to 97.8. And the trend is steep. CDC reports a 420 percent rise in ED visits for unsupervised melatonin ingestion by young children between 2009 and 2020, and notes that adult melatonin use in the US quintupled from 0.4 percent in 1999 to 2000 to 2.1 percent in 2017 to 2018. More product in more homes, in forms that look like candy. The poison-center series for 2012 to 2021 is a separate dataset, carried by article 325.
One regulatory fact from the report does more work than any advice could. Child-resistant packaging is not required for melatonin products, CDC states, although manufacturers may choose to use it. A parent comparing two bottles has no legal baseline to rely on.
On whether melatonin helps children sleep, the most rigorous recent answer comes from the Danish Health Authority, whose guideline panel published two GRADE-based systematic reviews in EClinicalMedicine in 2023. The first covered eight randomized trials in 419 otherwise healthy children and adolescents aged 5 to 20 with chronic insomnia of no identified cause. Total sleep time rose by 30.33 minutes and sleep latency fell by 18.03 minutes, both by sleep diary, with no heterogeneity between studies. No other benefit appeared. Non-serious adverse events rose, relative risk 3.44, and not one of the eight studies reported any information on serious adverse events, which is missing data rather than a clean safety record. The panel rated certainty very low for sleep quality and low for daytime functioning, and its recommendation states that “Melatonin should never be the first choice of treatment for this particular population.”
The companion review covered 13 trials in 403 patients aged 2 to 20 whose insomnia came from other medical disorders, with autism and ADHD studies excluded. Sleep latency fell 14.88 minutes, total sleep time rose 18.97 minutes, nothing else improved, certainty was low and the panel called the safety data scarce. The Danish Health Authority and the Parker Institute, supported by the Oak Foundation, funded both.
Where the evidence runs out
Long-term safety is the thinnest part of this literature, and the gap is specific rather than vague.
A 2019 systematic review in CNS Drugs by Frank Besag and colleagues examined adverse events across 37 randomized trials, at daily doses from 0.15 mg to 12 mg. Monitoring ran up to 29 weeks, but most studies lasted four weeks or less. The most frequent adverse events were daytime sleepiness at 1.66 percent, then headache, other sleep-related events and dizziness at 0.74 percent each and hypothermia at 0.62 percent. Few were serious. The reviewers also graded the studies: only eight of 37 were judged good against Cochrane risk-of-bias criteria, 16 fair and 13 poor. Their conclusion was that the shortage of long-term trials limits what can be said about continuous use, and that the data are too thin to assess concerns about at-risk groups.
Tolerance is undocumented in a trial literature made mostly of four-week studies, which is not the same as ruled out. What the review does report is that most adverse events resolved on their own within a few days, or immediately once treatment stopped.
The reassuring long-term pediatric data that circulates widely needs a disclosure attached. It comes from a 2022 review in Expert Review of Clinical Pharmacology reporting that long-term use of a pediatric prolonged-release melatonin product produced fatigue in 6.3 percent, somnolence in 6.3 percent and mood swings in 4.2 percent, with no evidence of effects on height, body mass index, pubertal development, tolerance or withdrawal. Its sole author works in the research and development unit of the company that makes that product, and the data concern a prescription prolonged-release medicine rather than an over-the-counter gummy. The reassurance does not transfer.
The American Academy of Pediatrics treats the unresolved questions as unresolved. Its parent-facing page, updated 22 January 2026 and written by integrative pediatrician Anna Esparham, names growth and puberty as an open concern, warns that melatonin supplements may also contain serotonin or cannabidiol, and lists morning sleepiness and more nighttime urination as common side effects.
Melatonin is absorbed quickly and broken down in the liver mainly by the enzyme CYP1A2, with more than 80 percent leaving as an inactive metabolite in urine. Drugs acting on CYP1A2 slow that breakdown. Article 353 covers interaction mechanisms in detail.
Questions for a clinician, not for a label
Nothing here is medical advice, and nothing in this article tells you to start, stop or change anything. These are the questions that make an appointment productive.
Is this insomnia or a clock problem? Trouble falling asleep at 11pm is a different problem from reliably falling asleep at 3am and waking fine at 11am. The 2015 AASM guideline supports timed melatonin for the second pattern, not the first.
What hour, and measured against what? The timing in the circadian research is anchored to a laboratory measurement, not to bedtime.
Who else in the house can reach it? Child-resistant packaging is not required, and gummies accounted for 47.3 percent of unsupervised ingestion visits among children under 6.
Does anyone involved have a seizure disorder? Australia’s regulator specifically flagged increased seizure frequency as a concern in children with seizure disorders.
What else is being taken? The CYP1A2 pathway matters, and so does additive sedation.
Three situations call for professional input ahead of anything else. A child, because the most rigorous guideline available says melatonin should never be first and the growth and puberty questions are open. An older adult with dementia, because the 2015 AASM guideline names that group in a recommendation against. And a suspected circadian rhythm disorder, the one diagnosis where the endorsement exists, because it turns on timing somebody has to measure.
If a young child swallows melatonin unsupervised, that is a poison-center and emergency-care question, not a wait-and-see one. Article 359 covers how adverse events get reported.
The one thing to carry away
Melatonin is a clock signal sold in the sleep aisle, and nearly every confusing thing about it follows from that mismatch.
The professional guidance is not contradictory. It says no for insomnia and yes for circadian disorders, because those are different problems. The measured effect on ordinary sleep is about seven minutes. The number on the package has an uncertain relationship to the amount in the capsule, and the clock research found a small dose at the right hour matching a large dose at its own. The fastest-growing group of users has the thinnest safety data and the sweetest-tasting products.
Sources
- American Academy of Sleep Medicine. “Pharmacologic Treatment of Chronic Insomnia in Adults – Guideline at a Glance (2017 clinical practice guideline).” aasm.org
- Journal of Clinical Sleep Medicine / American Academy of Sleep Medicine (verified via PubMed). “Auger RR, Burgess HJ, Emens JS, et al. "Clinical Practice Guideline for the Treatment of Intrinsic Circadian Rhythm Sleep-Wake Disorders… An Update for 2015." J Clin Sleep Med, 2015. PubMed. DOI: 10.5664/jcsm.5100.” doi.org
- American Academy of Sleep Medicine. “Health Advisory: Melatonin Use in Children and Adolescents (adopted September 2022; page last modified 12 March 2026).” aasm.org
- Centers for Disease Control and Prevention, Morbidity and Mortality Weekly Report. “Freeman DI, Lind JN, Weidle NJ, et al. "Notes from the Field: Emergency Department Visits for Unsupervised Pediatric Melatonin Ingestion — United States, 2019–2022." MMWR, 7 March 2024.” cdc.gov
- Therapeutic Goods Administration (Australia). “Safety concerns over imported counterfeit melatonin products (advisory updated 29 January 2026).” tga.gov.au
- European Medicines Agency. “Circadin (melatonin) – European Public Assessment Report.” ema.europa.eu
- American Academy of Pediatrics / HealthyChildren.org. “Esparham A. "Melatonin for Kids: What Parents Should Know About This Sleep Aid" (updated 22 January 2026).” healthychildren.org
- PLoS One (verified via PubMed). “Ferracioli-Oda E, Qawasmi A, Bloch MH. "Meta-analysis: melatonin for the treatment of primary sleep disorders." PLoS One, 2013. PubMed. DOI: 10.1371/journal.pone.0063773.” doi.org
- EClinicalMedicine / Danish Health Authority (verified via PubMed). “Edemann-Callesen H, et al. "Use of melatonin in children and adolescents with idiopathic chronic insomnia: a systematic review, meta-analysis, and clinical recommendation." EClinicalMedicine, 2023. PubMed. DOI: 10.1016/j.eclinm.2023.102048.” doi.org
- EClinicalMedicine / Danish Health Authority (verified via PubMed). “Edemann-Callesen H, et al. "Use of melatonin for children and adolescents with chronic insomnia attributable to disorders beyond indication: a systematic review, meta-analysis and clinical recommendation." EClinicalMedicine, 2023. PubMed. DOI: 10.1016/j.eclinm.2023.102049.” doi.org
- The Journal of Clinical Endocrinology & Metabolism (verified via PubMed). “Burgess HJ, Revell VL, Molina TA, Eastman CI. "Human phase response curves to three days of daily melatonin: 0.5 mg versus 3.0 mg." J Clin Endocrinol Metab, 2010. PubMed. DOI: 10.1210/jc.2009-2590.” doi.org
- CNS Drugs (verified via PubMed). “Besag FMC, Vasey MJ, Lao KSJ, Wong ICK. "Adverse Events Associated with Melatonin for the Treatment of Primary or Secondary Sleep Disorders: A Systematic Review." CNS Drugs, 2019. PubMed. DOI: 10.1007/s40263-019-00680-w.” doi.org
- Journal of Clinical Sleep Medicine (verified via PubMed). “Erland LAE, Saxena PK. "Melatonin Natural Health Products and Supplements: Presence of Serotonin and Significant Variability of Melatonin Content." J Clin Sleep Med, 2017. PubMed. DOI: 10.5664/jcsm.6462.” doi.org
- Expert Review of Clinical Pharmacology (verified via PubMed; author affiliated with a melatonin manufacturer). “Zisapel N. "Assessing the potential for drug interactions and long term safety of melatonin for the treatment of insomnia in children with autism spectrum disorder." Expert Rev Clin Pharmacol, 2022. PubMed. DOI: 10.1080/17512433.2022.2053520.” doi.org
- Sleep Medicine Reviews (verified via PubMed). “Yue J-L, et al. "Efficacy and tolerability of pharmacological treatments for insomnia in adults: A systematic review and network meta-analysis." Sleep Med Rev, 2023. PubMed. DOI: 10.1016/j.smrv.2023.101746.” doi.org