Walk into any drugstore and the melatonin choices are 3 mg, 5 mg, 10 mg, sometimes 20 mg in a berry-shaped gummy. Those doses are not sleep medicine dosing. They are marketing dosing. The physiological amount of melatonin the pineal gland secretes across an entire night peaks at plasma levels reachable with roughly 0.3 mg of oral supplementation. A single 10 mg gummy is roughly 30 times what the body would ever produce on its own.
Higher is not better here. It is not even the same drug. Melatonin at physiological doses is a circadian signal — a chemical message that says the body clock thinks it is nighttime. At the doses sold at retail, you get pharmacological levels that soak the receptors, produce daytime grogginess, and, by some measurements, work less well for sleep than the smaller dose does.
This piece is the honest read on melatonin dosing, timing, quality control, and who it actually helps. It is not a product recommendation. It is a case for using less of a thing that is already in short supply of solid published evidence.
What melatonin actually is
Melatonin is a hormone produced by the pineal gland in response to darkness. It is not a sedative. It does not put you to sleep the way an antihistamine or a benzodiazepine does. It signals to the body that biological night has started, and it plays a role in aligning circadian rhythm with the light-dark cycle.
That distinction matters. A sedative depresses central nervous system activity. Melatonin shifts or reinforces a circadian phase. The two feel similar at high doses because the massive pharmacological load produces some sedation as a side effect, but the mechanism is different. Confusing the two is why the retail category is a mess.
The Zhdanova 2001 dose-response study
The clearest single piece of evidence on melatonin dosing came out of Irina Zhdanova's lab at MIT. In a 2001 study in the Journal of Clinical Endocrinology and Metabolism, healthy adults received 0.1, 0.3, or 3.0 mg of melatonin or placebo, and were then tested on sleep onset in a nap protocol.
The 0.3 mg dose produced plasma melatonin levels close to the natural nocturnal peak and consistently shortened sleep latency. The 0.1 mg dose was too small to reliably reach physiological levels. The 3.0 mg dose reached plasma levels roughly 10 times physiological — it worked, but it produced hypothermia and residual sleepiness the following morning, and the sleep effect was not larger than the 0.3 mg dose. Zhdanova's conclusion, in plain terms: use the smallest dose that reaches physiological plasma levels. That dose is around 0.3 mg.
Nothing in the past two decades has overturned that finding. The doses at retail are 10 to 30 times what the evidence supports.
Why more is not more — Wurtman and receptor desensitization
Richard Wurtman, also at MIT, spent decades working on melatonin biology. One of his core observations was that melatonin receptors are subject to desensitization at supraphysiological doses. Flooding MT1 and MT2 receptors with 3 to 10 mg of exogenous melatonin can, over time, downregulate the receptors and blunt the response.
The practical version: taking a 10 mg gummy every night could be quietly making your own melatonin signaling less effective. The subjective feeling — "I need melatonin to sleep now" — has an obvious biological explanation. High doses are not building your sleep. They are training a downstream receptor to shrug.
Contrast that with the 0.3 mg physiological dose, which does not saturate receptors and does not produce measurable desensitization on repeat use in the published protocols. The evidence-based dose is small. The marketed dose is big. The gap between them is not accidental — a bigger dose feels stronger and sells better, even if it is worse for the actual mechanism.
What the meta-analyses show — and do not show
The Buscemi 2005 meta-analysis, still one of the more influential summaries of melatonin for sleep, pooled randomized trials across several sleep disorders. The pooled effect on sleep latency was a reduction of about 7 minutes. That is real, but small. For total sleep time, the pooled effect was about 8 minutes. The effect on subjective sleep quality was not statistically significant.
Two things about that meta-analysis matter for dosing. First, higher doses did not produce larger effects — the dose-response was flat or inverted upward. Second, the effect size does not remotely justify the "sleep aid" positioning at retail. A 7-minute reduction in sleep onset is worth having, but it is not a treatment for insomnia. Anyone with real chronic insomnia needs cognitive behavioral therapy for insomnia (CBT-I) — the actual first-line treatment — not a gummy.
Later reviews have refined the picture but not changed the shape. Melatonin has modest sleep-latency effects in healthy adults, larger effects in circadian rhythm disorders like delayed sleep phase and jet lag, and does not meaningfully treat classical primary insomnia.
The Erland and Saxena quality problem
Even if the science supported the 5 to 10 mg dose — and it does not — there is a bigger problem with retail melatonin: you often do not know what is in the bottle.
Erland and Saxena's 2017 analysis in the Journal of Clinical Sleep Medicine tested 31 commercial melatonin products. The finding: 71 percent of the products differed from labeled content by more than 10 percent. Some contained up to 4 times the labeled dose. Different lots of the same product varied. And, notably, roughly a quarter of the tested products contained serotonin as a contaminant — a substance with real drug-drug interaction concerns.
This is not a fringe finding. It is the reference point for understanding retail melatonin quality. In the United States, melatonin is regulated as a dietary supplement, meaning the FDA does not verify label accuracy pre-market. The best-quality products carry independent third-party certification — USP Verified or NSF Certified are the two most credible marks. Most retail melatonin has neither.
Chronobiotic vs sedative — two uses, two doses, two timings
The dose that works depends on why you are taking it. The two use cases are, functionally, two different drugs inside the same molecule.
As a chronobiotic (circadian shifter). This is the strongest evidence base for melatonin. If you have a delayed sleep phase — cannot fall asleep until 3 AM and wake at 11 AM, but want to be on an 11 PM to 7 AM schedule — a small dose (0.3 to 0.5 mg) taken 5 to 6 hours before your target sleep time can advance the phase over about a week. Same for eastward jet lag and for shift-work adaptation. The dose is small, the timing is early, and the mechanism is signaling to the body clock, not producing sedation.
As a sedative (30 minutes pre-bed). This is the retail use case. Higher doses taken 30 minutes before bed, aiming for a subjective sedation effect. The evidence for this use is the Buscemi-style 7-minute latency benefit, and the trade-off is receptor desensitization, morning grogginess, and the quality-control issues above.
If you are going to try melatonin at all, the chronobiotic protocol at 0.3 to 0.5 mg is a better bet than the 5 mg gummy 30 minutes pre-bed. But most people trying melatonin for "insomnia" do not have a circadian phase issue at all — they have anxiety, stress, poor sleep environment, or fragmented sleep architecture. Melatonin will not fix any of those.
Contraindications worth naming
Melatonin sits at the boundary of endocrine signaling and immune function. There are populations for whom the trade-off is worse than the small benefit.
- Pregnancy. Melatonin crosses the placenta. Human safety data on supplemental melatonin in pregnancy is thin. Physiological maternal melatonin is important for fetal circadian development, and adding pharmacological doses is not something the evidence base supports.
- Autoimmune conditions. Melatonin has immunomodulatory effects. For most autoimmune conditions the direction is not benign, and clinical guidance is generally to avoid supplementation.
- Certain psychiatric medications. Interactions with SSRIs, MAOIs, and benzodiazepines are documented but poorly characterized. Anyone on daily psychiatric medication should ask their prescriber before adding melatonin, not add it and see.
- Anticoagulants. Some evidence of interaction with warfarin. Uncommon but worth naming.
- Seizure disorders. Mixed evidence, some case reports of increased seizure activity. Talk to your neurologist.
None of these categories appear on the drugstore gummy label. That is a problem with the category, not with any individual product.
Children — where the mismatch is worst
Pediatric melatonin sales in the United States have exploded in the past decade. Emergency department calls for accidental pediatric melatonin ingestion rose by more than 500 percent between 2012 and 2021, according to CDC data — melatonin is now the most common substance in pediatric accidental ingestions.
The evidence base for melatonin in children is thinner than the retail volume suggests. There is decent evidence for melatonin in children with autism spectrum disorder or ADHD who have sleep-onset difficulties — the American Academy of Sleep Medicine has clinical guidance here. There is much weaker evidence for typically developing children with garden-variety sleep resistance. Long-term safety data in children is essentially absent. The pineal gland and hormonal signaling develop through adolescence, and there is a legitimate physiological worry about chronic pharmacological melatonin dosing in growing bodies — a worry pediatric endocrinologists have named publicly, without a definitive answer.
The honest position on kids: melatonin is not candy. If a child has a real sleep problem, the first-line moves are consistent bedtimes, a dark room, screens off, and pediatrician consultation. If melatonin is being used, use the smallest dose that works — 0.5 mg often is enough — and use it under pediatric guidance, not as a nightly retail gummy for years.
Test whether it is actually helping
The problem with any sleep aid is that subjective sleep quality is a bad measurement. People routinely report better sleep on nights when polysomnography shows no improvement, and vice versa. If you are going to try melatonin at any dose, get objective data on whether it is doing something for you.
A non-contact sleep mat placed under the mattress records movement, breathing, and heart rate through the night and estimates sleep latency, total sleep time, and wake-after-sleep-onset. It will not tell you a full polysomnography story, but for the specific question "did adding melatonin move the numbers," it is well suited. Run a two-week baseline with no melatonin, then two weeks with the intervention. Compare weekly averages, not single nights.
The bottom line on dosing
The evidence-based summary of a category that markets itself in doses 10 to 30 times too large:
- Physiological dose is around 0.3 mg. Zhdanova 2001 remains the reference point. Larger doses do not produce larger sleep effects and may desensitize receptors.
- Retail doses of 3 to 10 mg exist for marketing reasons, not clinical ones. If you must use retail sizes, cut them.
- Timing matters more than dose. Chronobiotic use is 5 to 6 hours before target sleep, small dose. Sedative use is 30 minutes before, and the evidence for it is weak.
- Quality control is a real problem. USP Verified or NSF Certified marks are the only real signals on label accuracy. Most retail products carry neither.
- Melatonin is not a sleeping pill. It is a circadian signal. Chronic insomnia not caused by circadian phase issues does not respond to it in any meaningful way.
- Children are not small adults. Talk to a pediatrician. Do not use gummies as a nightly routine.
Melatonin is a useful tool for a narrow use case — resetting a circadian clock that has drifted out of alignment. That is not the use case being sold in the sleep aid aisle. The mismatch between the evidence and the marketing is why so many people take melatonin, feel like it stopped working, and switch to something stronger. The answer is not stronger. It is smaller, earlier, and used for the actual problem in front of you.
