Melatonin
MT1 · MT2

Melatonin is the messenger that tells the body night has fallen and, above all, what time it is. It is a hormone produced by the pineal gland when the retina detects darkness, and from there it acts as a clock that tells the rest of the body when nighttime functions are due and when daytime ones are.

Melatonin is not a sleeping pill. What it does is more like setting the time than anesthetizing sleep, and that is why it works better if the problem is when you feel sleepy.

A substance able to “set” the biological clock has a technical name: chronobiotic. Melatonin acts on two receptors — MT1 and MT2 — located in the suprachiasmatic nucleus, the region of the hypothalamus that works as the body’s master clock. MT1 is more closely tied to inducing drowsiness; MT2, to shifting the clock’s phase forward or backward. That dual pathway explains why melatonin can, depending on how and when it is taken, help you fall asleep earlier or delay your internal clock by several hours.

The marker researchers use to know where your internal clock is is called DLMO (dim light melatonin onset, the start of melatonin secretion under dim light). Almost everything that follows depends on it: the ideal dose and timing are not fixed, but depend on where your DLMO sits relative to the time you want to sleep.

Biochemically it is N-acetyl-5-methoxytryptamine, synthesized in the pineal gland from serotonin. Nighttime secretion is controlled by the suprachiasmatic nucleus (SCN) through a pathway that begins in the retina —specifically in ganglion cells sensitive to blue light (~480 nm, via melanopsin)— and reaches the pineal gland through the retinohypothalamic tract and the superior cervical ganglion. The operational DLMO threshold is set at 3 pg/mL in saliva or 10 pg/mL in plasma.

The phase response curve (PRC) for melatonin is offset by about 12 hours relative to the PRC for light: taking it in the late afternoon or evening, before DLMO, advances the clock; taking it on waking delays it. This timing logic —not the dose— is what structures the dosing block and the evidence block further on.

Its most solid evidence is in timing problems: jet lag when crossing several time zones eastward, delayed phase disorder (falling asleep in the small hours even when you want to sleep earlier), and the misalignment of shift workers. In common insomnia the effect exists, but is small: it trims 4 to 7 minutes off how long you take to fall asleep.

Eastward jet lag is the best-studied case and where it helps most: traveling east forces the clock to advance, and the human clock (whose natural period is around 24.2 hours) is predisposed to the opposite, to delay. A Cochrane review of ten controlled trials concluded that melatonin is markedly effective at preventing or reducing jet lag, especially on trips of five time zones or more eastward.

For the traveling athlete, melatonin does not directly improve strength or maximal oxygen uptake; what it does is help resynchronize sleep. A trial in athletes who flew ten time zones eastward and took 5 mg over the following four days (Edwards et al., 2000, Ergonomics, PMID 11083138) recorded less fatigue and more alertness in the afternoon versus placebo.

In primary insomnia in adults (without an underlying timing disorder), according to several meta-analyses used here only as context and never as primary evidence, the effect ranges between 4 and 7 minutes less latency versus placebo. It is a real but small reduction. The exception is adults aged 55 or older, where prolonged-release 2 mg (the authorized medicine Circadin®) showed reductions of between 9 and 19 minutes, more marked from age 65, according to at least 5 pivotal trials, but it is worth knowing, and it is developed in the evidence section, that almost all of those trials are funded by the patent holder.

With melatonin there is no evidence that it improves memory, attention, or mental performance in someone whose sleep is well synchronized.

Delayed sleep-wake phase disorder (DSWPD) has its modern reference trial in Sletten et al. (2018), described in detail in the evidence section. The figure that qualifies its interpretation: in the subgroup where DLMO was measured after treatment, it did not change significantly, which suggests that much of the benefit comes from the soporific effect and the accompanying behavioral guidance, not from a real, measurable phase shift.

Indirect cognitive improvement via resynchronization has its clearest illustration in shift workers: a trial with 5 mg of nighttime melatonin after the shift, for four weeks, recorded a significant improvement on the occupational cognitive failures questionnaire in the fourth week versus placebo (Khanjani, Shamabadi, Akhondzadeh and Malekirad, 2024, Journal of Clinical Pharmacy and Therapeutics, DOI: 10.1155/2024/8869707). The cognitive effect appears when there is desynchronization to correct, not as a direct action on the awake, well-synchronized brain.

It makes the most sense for you if you fly frequently across time zones, work rotating shifts, or struggle to fall asleep before 2 or 3 in the morning no matter how hard you try. Outside those timing patterns, the benefit you can expect is more modest.

Important: the use in children, including for insomnia within autism (where there is solid clinical evidence), must always be prescribed and supervised by a healthcare professional, never a decision parents make on their own.

The American Academy of Sleep Medicine guidelines (Auger et al., 2015, Journal of Clinical Sleep Medicine, PMID 26414986) recommend, with low-to-moderate certainty on the GRADE scale, timed melatonin in DSWPD in adults and children, in pediatric irregular sleep-wake disorder, and in the non-24-hour rhythm of totally blind people. They explicitly recommend against its use in older people with dementia who have that same irregular disorder. They do not recommend it in advanced phase disorder or in the non-24-hour rhythm of those who retain vision.

Less is usually more. A small dose, similar to what your body produces naturally, performs as well as or better than one ten times larger, and without the side effect that does appear with the high dose: melatonin keeps circulating in the blood well into the morning and can leave residual drowsiness.

Timing matters more than the amount. To advance the clock (eastward jet lag, a delayed internal clock), the studied time is the afternoon, several hours before the desired bedtime. To delay the clock (westward jet lag), it is shortly after waking, not at night.

The research underpinning the “less is more” axis is from Zhdanova et al. (2001, Journal of Clinical Endocrinology & Metabolism, DOI: 10.1210/jcem.86.10.7901, PMID 11600532), in adults aged 50 or older. The physiological dose of 0.3 mg restored sleep efficiency without inducing hypothermia or keeping melatonin elevated during daylight hours, something that did happen with the 3 mg dose. The underlying explanation is that the body already has a well-calibrated system for producing melatonin in the amounts it needs; adding much higher doses does not perfect that system, it floods it for more hours than is desirable.

That timing matters so much has an uncomfortable side: taking it at the wrong time not only stops helping, it can make things worse. In a trial with 52 cabin crew members (Petrie et al., 1993, Biological Psychiatry, PMID 8513037), those who took it before the flight ended up worse than with placebo; those who took it after arriving improved.

In adults aged 55 or older with insomnia, the prolonged-release formulation (2 mg, released gradually over 8 to 10 hours, mimicking the natural profile) is taken one to two hours before bed, and is authorized by the European Medicines Agency for up to 13 weeks of continuous use in that indication.

What explains “less is more” pharmacokinetically is the magnitude of the plasma peak according to the dose: with 0.3 mg it is around 220 pg/mL, versus about 1,370 pg/mL with 3 mg (Zhdanova et al., 2001), an order-of-magnitude difference that leaves melatonin circulating far longer when the dose is high.

A significant part of the melatonin you take by mouth is lost before it reaches the blood: the liver processes and removes a good portion of it on its first pass. That is the reason the usual oral doses (1 mg or more) are higher than the amount your body produces naturally in a night.

The oral bioavailability is around 15%, with a wide range (9-33%) depending on the study and the person; first-pass (hepatic) metabolism removes about 85% of the dose. For that reason two people with the same pill can notice different effects.

There are presentations that avoid part of that first pass —sublingual or spray— absorbed directly through the lining of the mouth, with a faster onset of effect and greater systemic bioavailability than the oral route.

The main metabolism goes through the liver enzyme CYP1A2 (with CYP1A1 and CYP2C19 as secondary pathways), generating 6-sulfatoxymelatonin, which is eliminated in urine and accounts for about 89% of the administered dose. The half-life in immediate-release is about 45 minutes; in the 2 mg prolonged-release formulation (Circadin), the plasma peak is reached at 0.75 hours on an empty stomach, with a terminal half-life of 3.5 to 4 hours and a sustained release of 8 to 10 hours that reasonably mimics the natural nighttime secretion profile. In older people, the area under the plasma curve and the peak concentration are roughly double those in young adults, a relevant fact when thinking about dosing in that age range.

No, and here melatonin has a serious problem: what the label says and what is inside the bottle do not always match, and sometimes there is a lot of difference. An analysis of commercial products (Erland and Saxena, 2017) found actual contents ranging from 83% below the stated amount to 478% above, with more than 7 out of 10 products outside a reasonable margin (±10%) relative to their label.

This variability is not an isolated case or one particular brand with bad luck: it is a documented, recurring pattern in the OTC (over-the-counter) melatonin supplement market, made worse because, being a food supplement in most countries, it is not required to undergo the same prior content verification as a medicine. The problem is especially concentrated in chewable formulations, precisely the ones that, by their candy-like look and taste, are also those that cause the most accidental ingestions in children (developed in the evidence section and in contraindications).

The two reference analyses are Erland and Saxena (2017, Journal of Clinical Sleep Medicine, DOI: 10.5664/jcsm.6462), on 31 commercial products analyzed by mass spectrometry, which also detected serotonin as a contaminant in 26% of the products at potentially relevant doses; and Cohen et al. (2023, JAMA, DOI: 10.1001/jama.2023.2296), on 25 gummies, with the already-mentioned finding of the product with no detectable melatonin but with 31.3 mg of CBD. A third analysis, by the FDA, on 110 products aimed at children found a range of between 0% and 667% of the stated content, the most extreme figure of the three, though pending further independent verification, being a more recent and as-yet less-replicated finding.

USP Verified or NSF check that the actual content matches what is declared (seals uncommon in Europe). In the sports context, where avoiding banned substances also matters, examples of that kind of seal are Informed Sport or NSF Certified for Sport.

On its status in sport: melatonin is not included in the 2025 Prohibited List of the World Anti-Doping Agency (WADA), neither in nor out of competition. It is worth checking each season, since the list is reviewed annually, but as of this guide there is no restriction.

A selection of randomized, double-blind, placebo-controlled trials, the standard this site requires. The dossier brings together 25 main trials, of which only a few are later than 2020: melatonin is, paradoxically, one of the worst-funded supplements for new research despite decades of use, so much of the reference evidence is already fifteen or twenty years old.

The phase response curve (PRC) that underpins all the timing logic of this guide has its quantitative characterization in Burgess, Revell and Eastman (2008, Journal of Physiology, PMID 18006583) and, above all, in Burgess, Revell, Molina and Eastman (2010, Journal of Clinical Endocrinology & Metabolism, PMID 20410229), which directly compared 0.5 mg versus 3.0 mg: the maximum phase advance with 0.5 mg occurred when taken between 2 and 4 hours before DLMO (between 9 and 11 hours before the sleep midpoint); with 3 mg, the optimum shifts to about 5 hours before DLMO. There is also a dead zone of less than half an hour around DLMO itself in which there is barely any phase shift in either direction.

In shift work, besides the trial by Khanjani and colleagues already cited in the cognition section, several older trials with small samples (Sharkey et al., 2001; Sadeghniiat-Haghighi et al., 2008, with 86 shift-working nurses; Yazdi et al., 2016) agree on a modest but consistent improvement in total sleep time and latency, on the order of 15 to 25 minutes versus placebo, in those who sleep by day after a night shift.

Two recent clinical guidelines read the same body of evidence differently: the 2017 US one leans, with a weak recommendation, toward not using melatonin for insomnia in adults; the 2023 European one allows prolonged-release for up to three months in those over 55, with a moderate recommendation. They weigh the same finding differently.

All the evidence that follows is presented for its scientific value, not as an invitation to pediatric self-medication. Melatonin in children must always be the decision and follow-up of a healthcare professional.

It is worth qualifying the picture above with independent academic trials, without Neurim funding, that point in the same direction with smaller samples: Wright et al. (2011, Journal of Autism and Developmental Disorders) and Wirojanan et al. (2009, Journal of Clinical Sleep Medicine) found sleep-latency reductions of between 28 and 47 minutes versus placebo, with no declared industry ties. A more recent Japanese trial (Hayashi et al., 2022, Journal of Autism and Developmental Disorders) with 196 children, funded by the local manufacturer Nobelpharma, found a clear dose-dependent pattern: reductions of 22 minutes with 1 mg and 28 minutes with 4 mg, versus only 5 minutes with placebo.

Outside the circadian domain, the only emerging indication with reasonably solid support, a Cochrane meta-analysis of moderate certainty, cited here only as secondary context, is preoperative anxiety, where melatonin showed an effect comparable to midazolam in reducing anxiety measured before surgery (Madsen et al., 2020, Cochrane Database of Systematic Reviews). The rest of the emerging uses (cluster headache, migraine, delirium in intensive care) have mixed, preliminary, or outright inconsistent evidence across studies, and do not reach the level of solidity of the above.

The figure from Lelak et al. is verified and exact: 260,435 total ingestions over the period (not 530,000 —that figure corresponds to the percentage increase, not the total), with a particularly marked rise between 2019 and 2020. A later CDC analysis (Freeman et al., 2024, MMWR, DOI: 10.15585/mmwr.mm7309a5) attributes some 11,000 additional emergency room visits to unsupervised melatonin ingestion between 2019 and 2022. A prevalence study (Hartstein et al., 2024, JAMA Pediatrics, DOI: 10.1001/jamapediatrics.2023.4749) puts melatonin use in the past month at around 5.6% of preschoolers and nearly 20% of school-age children in the United States —a background exposure figure that helps make sense of the scale of the accidental-access problem. The American Academy of Sleep Medicine explicitly recommends treating melatonin, in the pediatric context, as a medicine and not as a harmless product, and always consulting a professional before giving it to a child.

There is no direct comparison between immediate- and prolonged-release in insomnia of older adults. Each formulation has its own trials, but no study has pitted them head-to-head in the same population to find out which works better in practice.

The effect of chronic exogenous melatonin on puberty is an open question with no direct answer in humans. In animal models, melatonin suppresses gonadotropin-releasing hormone (GnRH), which in theory could delay pubertal development; an ethically feasible trial would only exist if it were carried out over years in children, which raises obvious design problems. The longest follow-up available in children with autism and prolonged melatonin use (close to seven years) suggested a subjective tendency toward some pubertal delay, without it being objectively confirmed. It is, honestly, the most relevant uncertainty in this whole guide for prolonged pediatric use, and it justifies the active clinical monitoring already mentioned in the contraindications section.

Nor is there a randomized, double-blind, adequately sized trial in the non-24-hour rhythm of blind people after the pivotal study of the year 2000 —it remains, with its seven participants, the best data available almost 25 years later, simply because the disorder is rare and no one has repeated the trial on a larger scale.

The immunological effects of melatonin have only heterogeneous preclinical data, with no solid clinical studies in humans that would allow a conclusion to be drawn.

The real optimal dose still lacks consensus. Laboratory evidence points to 0.3 mg as the reference physiological dose, but the market mostly sells between 1 and 10 mg, with no formal clinical consensus on what the minimum effective dose is for each use.

In the short term, in healthy adults, melatonin has a good safety profile. Even so, there are situations in which it is wise to talk to a doctor before taking it and one, above all, in which the decision should never be made at home without supervision:

• Children and adolescents. However many over-the-counter products with a harmless appearance exist, melatonin in minors is not self-care. Its pediatric use, including in the autism context where there is solid clinical evidence, must always be prescribed and supervised by a healthcare professional.
• Pregnancy and breastfeeding. It is not recommended, due to a lack of sufficient clinical data —not because firm proof of risk exists, but because the evidence needed to rule it out with confidence is not there.
• Treatment with fluvoxamine (an antidepressant). This combination is documented and is one of the most relevant interactions in this whole guide: fluvoxamine can multiply melatonin blood levels by 17. While on fluvoxamine, medical assessment before adding melatonin is necessary.
• Driving and operating machinery. The effect can persist for several hours; the usual caution period is the 4-5 hours after taking it.

As with any supplement, the above is information, not a substitute for a doctor’s assessment, especially with other ongoing medication or a diagnosed chronic illness.

There is also a group of situations where clinical prudence recommends prior consultation, even if the evidence of direct risk is weaker or less studied: autoimmune diseases (melatonin has an immunomodulatory role that is not fully characterized), hepatic insufficiency (its metabolism is hepatic), epilepsy (the risk signal is weak and the data contradictory, but it exists), and coagulation disorders or treatment with anticoagulants such as warfarin, where it is advisable to monitor the INR if combining them under medical supervision.

Drug interactions to keep in mind (information, not a substitute for medical assessment in the case of treatment with any of these drugs):

On pregnancy and breastfeeding, the summary of product characteristics of the prolonged-release medicine authorized in Europe does not recommend it, due to the absence of clinical data in humans, drawing on animal studies that do show placental transfer and passage through milk.

A curiosity worth knowing because it changes how its “safety” is perceived: the same molecule has completely different statuses depending on the country. In the United States it is bought over the counter in any supermarket; in the United Kingdom and Japan it is a medicine dispensed only with a doctor’s prescription. In the European Union it is, in general, a food supplement with dose limits that vary widely from one country to another. That disparity does not reflect differences in the available scientific evidence —it is the same in every country— but different regulatory decisions about where to set the threshold of “medicine” versus “supplement.”

Two specific medicines underpin much of the European regulatory evidence cited in this guide: Circadin (2 mg prolonged-release, authorized in 2007 for insomnia in those over 55) and Slenyto (authorized in 2018 for insomnia in children with autism by prescription, with subsequent extensions to other developmental disorders). Both are, as noted in the evidence section, owned by the same pharmaceutical company.