Why Doesn't Melatonin Work?

Last checked 29 October 2026 · 13 min read

Because it is not a sleeping pill, and it is being asked to do a sleeping pill's job. Melatonin tells the body what time it is. It does not switch consciousness off, and it cannot quiet a mind that is already in bed at the right hour. So it helps a clock in the wrong place — jet lag, shift work, a night that starts three hours too late — and does close to nothing for the insomnia most people are actually complaining about. Less widely repeated: the same dose that shortens sleep onset at eight in the evening can lengthen it at nine, because by then your own melatonin has already risen and there is nothing left to move. Timing fails more often than dose does.

It is a clock signal, not a switch

Melatonin is a hormone the pineal gland releases in the evening, in darkness, and stops releasing when light reaches the retina in the morning. Its job is not to make you unconscious. Its job is to tell every tissue in the body what time it is, so that temperature, digestion, alertness and repair all happen in the right order. Sleep follows that signal the way a train follows a timetable: the timetable does not pull the train, but nothing runs on time without one.

Two consequences follow from that, and almost every disappointment with melatonin is one of them.

A correct signal added to a correct clock does nothing. If you are in bed at eleven, and your body also believes it is eleven, then your melatonin is already where it should be. Taking more of a signal you are already producing at the right moment is like telling someone the time when their watch is accurate. In one controlled trial, people who slept normally were given three different doses of melatonin across three weeks — and none of the doses changed their sleep at all. The people whose sleep improved were the ones whose sleep was already broken.

The dose is small on purpose. A supplement is not a sedative and does not behave like one. It does not push you under; it lowers the threshold at which sleep becomes possible, and only if the rest of the system is pointed the same way. This is why the honest description of what melatonin does is boring: it makes the evening feel like evening. If your evening already feels like evening and you are still staring at the ceiling, the hormone was not the missing piece.

The same pill, three hours apart

Here is the finding that explains most of the contradictions in the accounts people give. In a small laboratory study, six healthy young men took either 0.3 mg of melatonin, 1.0 mg, or a placebo, at three fixed times: six in the evening, eight, and nine. They were wired up and allowed to sleep immediately afterwards.

At six and at eight, melatonin shortened the time it took them to fall asleep. At nine, the same 0.3 mg dose made sleep onset slower, and the 1.0 mg dose did nothing at all. Nothing about the pill had changed between eight and nine. Only the hour had.

There are two ways to read that, and they are not mutually exclusive. The first is a ceiling: by nine, the body's own melatonin has begun to rise, the receptors are already occupied, and an extra dose has nothing to bind to. The second is direction: melatonin does not only say night, it says later or earlier depending on when it arrives relative to your internal clock, and a dose that lands on the wrong side of that rise can nudge the clock the wrong way. Which of the two is doing the work in any given person has not been settled. Both predict the same thing for you, which is the useful part: past a certain hour, more melatonin stops helping and starts costing.

This is also why the standard instruction — take it thirty minutes before bed — is not wrong everywhere, and is the instruction most likely to fail you. In the study of older insomniacs described further down, thirty minutes before bed worked. Those people were not late; they were shallow, and the dose was replacing something they had stopped making enough of. If your problem is that your night starts too late, thirty minutes is already too late to move it.

The failures that are not failures

Most people who say melatonin did nothing are describing one of five things, and four of them are not evidence about the hormone at all.

They took it once. A clock is not reset in a night. Shifting a body's timing takes consecutive evenings at the same hour, because each dose moves the internal clock a little and the movement accumulates. A single tablet before one bad night is a test of nothing.

They took it too late. Covered above, and it is the most common of the five. Thirty minutes before bed is a sedative's schedule, not a clock signal's schedule. If the goal is to move the hour you fall asleep, the dose has to arrive two to four hours before the bedtime you want, while the clock is still movable.

They took a dose chosen by the packaging. Gummies are sold in 5 mg and 10 mg — more than ten times the dose that restored sleep in the trial described below, and far above anything the body produces in a night. The pill bottle is a worse guide to dose than the research is, and it is the only guide most people have.

Their problem was not the clock. This is the big one, and it gets its own table below. If the difficulty is at the start of the night — wired, alert, thinking — melatonin is being asked to fix arousal, and arousal is not a timing problem. Wired but tired is the closer page.

The room was brighter than the dose. Light reaching the eye in the evening suppresses the body's own melatonin, and a tablet is a small addition to a system that is being actively told it is still daytime. Taking it with a phone at arm's length and every lamp on is not a fair trial of anything.

Why it can make you more awake

This is the part that no guide prepares people for, and it is real: for some people melatonin does the opposite of what the label promises. Three accounts, all of them specific, and none of them matching.

The ten-milligram gummy. Someone with a year of insomnia took melatonin at a partner's suggestion and described the mind becoming hyperactive within the hour, with vivid imagery never experienced before and a sense of electrical activity running down the body. They were more awake than before the dose, and uncomfortable with it. They had not chosen the amount; the gummies came that way, and they had assumed that was normal.

The one milligram that cleared a head. Someone else took a quarter of a 4 mg tablet — one milligram — for five or six nights. It did not help them sleep; it made falling asleep harder. But it removed a headache, cleared a brain fog carried for months, made complex thinking easier, and reduced body pain. They were not sedated. They were better, in a way that had nothing to do with sleep, and they wanted to know whether they could take it in the daytime instead.

The dose that stopped working, then worked again smaller. A third person used 3 mg successfully for a while, then found it stopped helping and their sleep became unusually light. They opened a capsule and took a small amount of the powder under the tongue; sleep improved again. Meanwhile, in the same discussion, someone insisted the opposite — that only the higher doses helped them and the low doses produced exactly the wired, head-clearing effect described above.

The honest summary is that the direction of the effect varies between people and nobody can currently tell you which way you will go. What is consistent across all three accounts is that the effect depends on dose and on timing — which means the two variables worth changing are the two that people almost never change. If melatonin made you feel strange, the first experiment is not to give up on it. It is to cut the dose by three-quarters and move it two hours earlier.

Two side effects are worth naming because they are under-reported and unmistakable when they happen: unusually vivid dreams, and nightmares vivid enough to wake you. Several people describe both, and some describe sleep paralysis arriving with them. They are not dangerous and they usually fade, but they are a reason to stop rather than push through, particularly if the dreams are the reason you are waking.

The dose on the label is not the dose in the tablet

Before any of the above matters, there is a measurement problem. Melatonin is sold as a supplement in most countries, which means it is not held to the content standards of a medicine. When researchers analysed thirty-one products — different brands, tablets, capsules, liquids — the amount of melatonin inside ranged from 83% below the label to 478% above it. Batch-to-batch variation within a single product reached 465%. More than seven in ten products missed their own label by more than 10%. Eight of the thirty-one contained serotonin, at levels between 1 and 75 micrograms, which is not on the label at all.

What that means in practice: two people comparing notes about melatonin may not have been taking the same substance in any meaningful sense. "It did nothing for me" can mean the tablet contained a tenth of what it claimed; "it made me feel awful" can mean it contained five times what it claimed. Neither is a verdict on the hormone.

The workarounds are unglamorous. Buy a product that carries third-party testing — a USP mark, or an equivalent independent verification — because that is a statement about content rather than about marketing. Buy the lowest dose available and split it, rather than buying a high dose and hoping. Stay with one brand and one batch while you are testing, so that a change in your sleep is a change in you and not a change in the tablet.

What the numbers actually support

Two independent lines of evidence agree on the shape of the answer, which is unusual and worth stating plainly.

The clinical guidance is small and early. The medically reviewed consumer guidance in wide circulation recommends starting at 0.5 to 1 mg, thirty minutes before bed, and only moving up to 3 to 5 mg if that fails. It puts the peak of the benefit at around 4 mg, says outright that more than 4 to 5 mg is unlikely to add anything, and adds a detail most people skip: if it is not working, changing when you take it is more likely to help than increasing the dose.

The trials used amounts that would look like a mistake on a label. In a double-blind study of people over fifty with confirmed insomnia, three doses were tested against placebo, each for a week: 0.1 mg, 0.3 mg, and 3.0 mg, all taken thirty minutes before bed. The 0.3 mg dose — a physiological replacement, not a dose — restored sleep efficiency, working mainly in the middle third of the night, and brought nighttime melatonin levels back to normal. The 3.0 mg dose also improved sleep, but it lowered core body temperature and kept melatonin elevated into daylight hours. And the people in the same study who slept normally were unaffected by every dose.

Put those together and the picture is not "melatonin is weak". It is that melatonin has a narrow useful range, that the range is far below what is sold, and that above it the extra dose buys side effects rather than sleep. Almost everyone who reports that melatonin stopped working, or turned on them, is somewhere above that range.

Clock problem or arousal problem

This is the distinction that decides whether melatonin is the right tool, and it is answerable from your own week without buying anything.

Most people with chronic insomnia are in the right-hand column, which is why the supplement aisle keeps disappointing them. The two problems feel similar from inside — both are "I cannot sleep" — and they respond to opposite treatments. Melatonin on an arousal problem is a timetable handed to someone whose train will not start.

The order to check, if it has never worked

Work down this list rather than changing everything at once. Each step is a week, and each one changes a single variable.

When to stop rather than adjust

If it is giving you vivid dreams or nightmares that wake you. Not a reason to reduce; a reason to stop. The dreams tend to return each time you restart, and the sleep you are protecting is worth more than the experiment.

If your mornings got worse. A dose large enough to keep levels elevated into daylight produces exactly the flat, foggy morning that people then try to fix with coffee. If the mornings changed when the melatonin started, the melatonin is the change.

If you are pregnant or nursing. Supplementation is not recommended in either case, because the research that would make it safe has not been done.

If you take any of a long list of medicines. Melatonin interacts with sedatives and sleep aids, blood thinners, anticonvulsants, blood pressure medication, antidepressants, oral contraceptives, diabetes medication and immunosuppressants. This is a conversation with a pharmacist, not a label to read.

If you need it every night and have for months. Long-term safety has not been studied well enough for anyone to promise you it is fine, and needing a nightly tablet to start the night is itself a finding worth taking to a clinician. Not because the tablet is dangerous, but because of what the need is telling you.

Where this fits

If the difficulty is at the start of the night and your body feels switched on rather than sleepy, that is arousal, and wired but tired is the page that treats it as a lock rather than a shortage. If the effort of falling asleep has become the problem in itself, how to stop trying to sleep is the one that costs nothing. If the night breaks in the middle rather than at the start, that is a different mechanism entirely and waking at three is closer. And if the tiredness never turns into sleepiness at any hour, the distinction between the two is in tired but not sleepy.