Melatonin reduces sleep latency by an average of seven minutes and adds eight minutes of total sleep. That's the meta-analytic answer from 26 randomized trials.
Most over-the-counter melatonin in North America is sold at 5 or 10 milligrams. The dose actually associated with that seven-minute effect, in the studies that found it, is between 0.3 and 1 mg. People are taking ten to thirty times what the evidence supports — for an effect smaller than most expect.
I. What melatonin actually does
Melatonin is not a sleeping pill. It is a hormone the pineal gland produces in response to darkness, and its role in the body is to signal time, not enforce sleep. The technical word is zeitgeber — a German term sleep researchers borrow to describe an external or internal cue that anchors the circadian clock. Light is the strongest zeitgeber. Melatonin is the second strongest.
This distinction matters because it predicts what melatonin will and won't do.
It will:
- Help reset a circadian phase that is genuinely shifted (jet lag across multiple time zones, deliberate shift-work adjustment, delayed sleep phase syndrome).
- Modestly shorten sleep onset, particularly when taken at the right time and in physiological doses.
- Modestly extend total sleep time.
It will not:
- Sedate you in any pharmacological sense.
- Override caffeine, screen exposure, or anxiety.
- Reset a circadian phase that is not shifted — which is most insomnia.
Most consumer use of melatonin treats it as a sedative. That use case is largely unsupported by the literature, and the reason it "works" anecdotally is partly placebo, partly the small real effect, and partly the secondary effect of finally committing to a bedtime ritual.
II. The meta-analytic answer
In 2024, a team led by Francy Cruz-Sanabria published a comprehensive systematic review and dose-response meta-analysis in the Journal of Pineal Research (Wiley, 10.1111/jpi.12985). It pooled 26 randomized controlled trials with 1,689 participants total, examining time, dose, and outcome measures across the literature.
The headline numbers:
- Sleep onset latency: reduced by an average of ~7 minutes
- Total sleep time: increased by an average of ~8 minutes
- Dose-response peak: around 4 mg/day, beyond which additional dose did not produce additional effect
- Optimal timing: approximately 3 hours before desired bedtime, not the conventional "30 minutes before sleep"
Three observations are worth pausing on:
The effect is real but small. Seven minutes of latency improvement is statistically significant in pooled trials and clinically modest in everyday terms. It is the difference between falling asleep at 11:07 vs. 11:14 — not the difference between sleeping and not sleeping.
The dose ceiling is roughly 4 mg. Beyond that, the dose-response curve flattens. Taking 10 mg doesn't produce more sleep than 4 mg. It does, however, produce more side effects (vivid dreams, next-day grogginess, headaches) and possibly meaningful long-term costs (see Section IV).
The timing convention is wrong. The standard "take it 30 minutes before bed" advice was carried over from sedative pharmacology. Melatonin's mechanism is circadian-phase shifting, and that mechanism takes several hours to act. Three hours before desired bedtime aligns with the pharmacokinetics far better — and in trials, produces a stronger phase-shift effect.
III. The dose gap
A useful frame for the dose problem comes from comparing four numbers:
| Source | Dose |
|---|---|
| Endogenous nightly production by the pineal gland | ~0.3 mg |
| Effective dose for circadian-rhythm uses (jet lag, DSPS) | 0.3–1 mg |
| Dose-response peak in 2024 meta-analysis | ~4 mg |
| Common over-the-counter consumer doses (US) | 5–10 mg, often 12+ mg |
The body produces about 0.3 mg of melatonin per night under normal conditions. Doses in the 0.3–1 mg range, taken at the right time, mimic that endogenous signal. Doses in the 5–10 mg range deliver 17 to 33 times the body's natural production.
That overdosing isn't merely "more than needed." It produces sustained supraphysiological melatonin levels in circulation — including during the daytime hours after waking, when melatonin is normally near zero. The downstream consequences of long-term daytime melatonin elevation aren't fully characterized, but recent observational work suggests they are not negligible.
Why does the OTC market sell 10 mg pills, then? Two reasons. One: in the US, melatonin is sold as a dietary supplement, not a drug. There is no FDA dose ceiling, no required label accuracy, no standardized formulation. Independent assays have repeatedly found the actual content of OTC melatonin products varies by 80%+ from labeled dose, in both directions. Two: higher doses produce more subjective drowsiness, even if not more sleep, and consumers reward bottles that "feel like they work."
IV. Long-term safety: a recent flag
In November 2025, researchers presented preliminary results at the American Heart Association's Scientific Sessions analyzing five years of electronic health records from 130,828 adults with chronic insomnia. Adults using melatonin long-term were associated with a 3.5× higher rate of hospitalization for heart failure compared to non-users.
Several caveats apply, and they are important:
- The study is observational. Causation cannot be inferred from association. People who take melatonin long-term may differ from those who don't in ways that themselves predict heart failure (chronic insomnia itself is a cardiovascular risk factor; lifestyle factors correlated with chronic supplement use; etc.).
- The findings are preliminary, presented at a conference, not yet peer-reviewed or published in final form.
- The dose distribution in the study population is not specified — most US melatonin users take 5–10 mg, which is far higher than the doses examined in efficacy trials.
What the study does not establish: that melatonin causes heart failure. What it does suggest: long-term high-dose melatonin use, common in the US consumer population, is in a domain where the safety database is thinner than people assume, and where signals are starting to appear.
For the same period, no similar signal appears for low-dose, short-duration use (e.g., 0.3–1 mg for occasional jet lag). The risk profile of "use melatonin twice a year for international travel" is very different from "5 mg every night for two years."
V. The non-pharmacological alternative most people don't see
The framing throughout this essay has been "what melatonin does and doesn't do." But the deeper question — what works for chronic insomnia — has a different and better-evidenced answer that consumer marketing rarely surfaces.
Cognitive-behavioral therapy for insomnia (CBT-i) is the first-line treatment recommended by the American Academy of Sleep Medicine. It outperforms sleeping medications in long-term outcomes, and a 2025 meta-analysis of digital CBT-i (Lee et al., npj Digital Medicine, 49 RCTs, 20,000+ participants) found medium-to-large effect sizes that persist at 6-month follow-up — a duration over which medication effects typically fade.
Two specific components of CBT-i carry most of the effect:
Sleep restriction. Compressing time in bed to closely match actual sleep duration, then gradually expanding. This builds homeostatic sleep pressure (Process S, in the two-process model) and breaks the conditioned association between bed and wakefulness.
Stimulus control. Restricting bed use to sleep and intimacy only — no reading, no phone, no anxious lying-awake. After 15+ minutes of inability to sleep, get out of bed; only return when sleepy.
These principles are simple to describe and difficult to execute alone. Self-administering sleep restriction without coaching tends to fail at compliance, which is why a clinician-supported version (digital or in-person) outperforms self-help. But the principles themselves are public, well-evidenced, and worth knowing — they explain a lot about why some non-melatonin approaches work and most do not.
VI. What's next
The next essay (Insomnia has four phases — and 90% of advice only addresses one) is an extension of this point: not all insomnia is the same, and the "one-size-fits-all" model implicit in most consumer advice (including melatonin marketing) misses the most important variable in choosing a response.
A practical orientation for melatonin specifically: if you take it, consider 0.3–1 mg, three hours before desired bedtime, and treat it as a signaling tool rather than a sleeping aid. This is the dose and timing the meta-analysis supports. If you've been taking 5–10 mg for months and seeing no clear effect — there's a high chance you've been taking ten times too much for the wrong reason at the wrong time, and the bottle is no longer doing anything except making the drawer look full.