Name: Melatonin: A Clock Signal Worth Skipping for Most People
Creator: Realised
Published: 2026-10-01

Summary

Melatonin is a timing signal, not a sleeping pill — there is no sign that it suppresses your own production or causes dependence (the evidence against that fear is moderate, from one long-term study), but "low-risk" is not "zero downside for everyone": it shifts glucose tolerance, interacts with common drugs, is sold in the US at doses 74–347% of the label, and is sold only as a medicine, mostly on prescription, in the UK, Japan and Australia, while in the EU the rules vary and several countries allow low doses in food supplements. Its actual sleep benefit is tiny (~7 minutes faster onset), an

Why Strong

Tier 1 overall: the modest efficacy (Ferracioli-Oda; AASM guideline), the product-quality data (JAMA), and the regulatory facts are all Strong. The no-suppression finding (Lemoine) is Moderate: one open-label, maker-funded study that measured 15 people against published values. The glucose-tolerance downside is Tier 2 (gene-interaction trending Tier 1); the paediatric puberty concern is Tier 3 (concern) / Tier 4 (for any claimed effect) and labelled as such. The overall verdict is conditional, not blanket — which is the correction from the prior version.

NOT Tier 0.5 because optimal use involves judgement (dose/timing/indication) and a genuine, multi-sided controversy. NOT Tier 2 because the load-bearing facts for the verdict (modest effect, product/regulatory reality) are well-replicated.

Practical takeaway

For ordinary sleep trouble: fix the inputs, skip the pill. Morning daylight, dim/warm light in the last 1–2 hours before bed, consistent sleep–wake times, daytime activity (see circadian_rhythm_optimization, morning_sunlight_exposure). This is free, fixes the cause, and pays dividends melatonin can't. For chronic insomnia, the first-line treatment is CBT-I, not a supplement (see cbt_i_program_overview) — melatonin is not recommended for it.

If you supplement, treat it as a clock-setter for a genuine circadian problem (jet lag, shift work, delayed sleep phase):
• Dose low: 0.3–0.5mg (up to ~2mg prolonged-release). Most products are 10–30× this — and may contain far more than the label says.
• Time it early, not at lights-out: ~2–3 hours before target bedtime for phase-advance.
• Take it well away from food (≥2 hours): to avoid the glucose-tolerance hit — especially if you're diabetic, pre-diabetic, or know you carry the MTNR1B variant.
• Buy quality: third-party-tested (the US market is badly mislabelled); avoid gummies for dosing accuracy.
• Keep it short-term. Long-term safety at the high OTC doses isn't established.

Don't give it casually to children (see Risks) and don't use it as a nightly hypnotic for general insomnia.

Evidence detail

Why This Entry Exists

Three errors surround melatonin, not two. The supplement aisle says "take 10mg to knock yourself out." The wellness crowd says "never touch it, it shuts down your gland." And the reassuring take — the one this entry used to make — says "it's basically harmless, the only myth is the dependence one." All three overreach, and the third is the subtlest: debunking the suppression myth does not establish that melatonin is safe for everyone at every age, because the genuine downsides live somewhere else entirely (metabolism, children, drug interactions, product quality).

The honest picture is conditional. Melatonin is a hormone that tells your brain it's night — useful as a clock-setter for genuine circadian problems, largely pointless as a nightly sedative, and not free of risk. For the average person, the high-leverage move is upstream of the bottle: fix the light and timing inputs that raise your own melatonin, which fix the cause and pay dividends a pill never will.

What bad advice this protects against, in all directions:
• Megadose-as-sedative → 5–10mg nightly for ordinary insomnia, where the benefit is ~7 minutes and a body of guidelines advises against it.
• "Melatonin shuts down your gland / it's addictive" → avoiding a legitimate circadian tool on a mechanism the one long-term study that looked found no sign of.
• "It's a harmless supplement, give it to anyone including kids" → ignoring the metabolic, interaction, paediatric, and product-quality downsides — and the fact that most of the world regulates it as a medicine.
• Reaching for the bottle instead of the light → treating a symptom of circadian misalignment while the free, multi-benefit fix sits unused.

Evidence

1. It does NOT appear to suppress endogenous production (Tier 2) — the one genuine myth-debunk, resting on one study. An open-label long-term study funded by the maker of the tablet it tested (Lemoine et al. 2011, n=244, prolonged-release melatonin 6–12 months) measured nocturnal 6-sulfatoxymelatonin in 15 of the patients two weeks after they stopped a 6-month course, and found output similar to published values for people of the same age without insomnia: no sign of suppression of your own production. No demonstrated negative-feedback onto the pineal at supplement doses; no gland atrophy. Graded Moderate rather than Strong because this rests on one open-label, maker-funded study that measured only 15 people and compared them with published values rather than a control group.

2. It does NOT appear to cause dependence or rebound insomnia (Tier 2). Same study, so graded Moderate for the same reason as point 1: discontinuation produced no rebound insomnia and no withdrawal (CHESS-84 score 1.37 vs 7.06 for lorazepam). Melatonin is not habit-forming — a real contrast with Z-drugs and benzodiazepines. (These two findings are why the "it shuts your system down" fear fails — but they say nothing about the downsides below.)

3. The actual sleep benefit is small, and it's a chronobiotic not a hypnotic (Tier 1). Meta-analysis (Ferracioli-Oda 2013, 19 RCTs, n=1,683): melatonin cut sleep-onset latency by only ~7 minutes and improved sleep quality modestly (SMD +0.22) — smaller than other hypnotics. The **AASM clinical practice guideline (Sateia 2017) recommends against melatonin for chronic insomnia in adults**, while supporting strategically timed melatonin for jet lag, shift-work disorder, delayed sleep-wake phase, and non-24 in the blind. Its real value is shifting the clock, not sedation.

4. It shifts glucose tolerance — the biggest under-appreciated downside (Tier 2, gene-interaction trending Tier 1). Melatonin receptors on pancreatic β-cells suppress insulin secretion, so melatonin near food impairs glucose handling:
• Rubio-Sastre 2014 (Sleep): a single 5mg dose with a glucose load raised glucose AUC +186% in the morning, +54% in the evening vs placebo — hence "take it ≥2 hours after eating."
• Garaulet 2022 (Diabetes Care): the MTNR1B rs10830963 G-allele (carried by ~30–50% of people) magnifies it; whole-population late eating gave +8.3% glucose AUC, −6.7% insulin AUC.
This is a real physiological downside, entirely separate from the dependence myth — and it directly refutes "zero downside."

5. US product quality is genuinely bad (Tier 1). Erland & Saxena 2017 (J Clin Sleep Med): >71% of 31 supplements were outside ±10% of label; serotonin detected in 26%. Cohen 2023 (JAMA): 22 of 25 gummies (88%) mislabeled, 74–347% of label; one contained CBD and no melatonin. You frequently don't know the dose you're taking — which compounds every dose-dependent downside.

6. The regulatory split is itself evidence (Tier 1). Melatonin is prescription-only / restricted as a hormone-medicine in the UK (MHRA), Japan, and Australia (TGA) but an OTC dietary supplement in the US. In the EU the rules vary by country. The EU-wide medicine licences are prescription-only prolonged-release tablets (Circadin 2mg, adults 55+, short-term; Slenyto for children with autism, ADHD or certain genetic conditions), while several member states, among them France, Italy, Spain, Belgium and the Netherlands, also allow low doses (0.3 to 2mg a day) as food supplements. Several also sell immediate-release melatonin over the counter as a medicine. Multiple serious regulators gate it citing drug interactions and insufficient broad safety evidence. "Harmless supplement at any dose" is a US framing, not the global expert consensus: where EU countries allow melatonin in supplements, they cap it at 0.3 to 2mg a day.

7. Low timed doses are sufficient; megadoses are pointless (Tier 1–2). ~0.3–0.5mg taken 2–3 hours before habitual sleep acts as a chronobiotic; benefit plateaus around 4mg; 5–10mg gives supraphysiologic levels, no extra benefit, and more next-day grogginess.

Mechanism

What melatonin is. A hormone the pineal secretes after dark, rising ~2 hours before natural sleep onset (the dim-light melatonin onset, DLMO). It is the body's internal nightfall announcement — permissive, not sedating — which is why it never feels like a sleeping pill and why timing matters more than dose.

Why timing dominates dose. Melatonin shifts the clock via the suprachiasmatic nucleus along a phase-response curve: a small dose in the early evening advances the clock; the same dose at the wrong time does little. Dose mostly governs how supraphysiologic your blood level (and side-effect load) gets.

Why it touches glucose. Pancreatic β-cells carry melatonin (MT2) receptors that suppress insulin release. High melatonin + carbohydrate (an evening megadose with a late meal, or any dose near food) blunts insulin and raises post-meal glucose — markedly so in the ~30–50% carrying the MTNR1B risk allele.

Why "it shuts down your gland" does not hold up. No demonstrated pineal negative-feedback at supplement doses; in the one long-term study that measured it, own output looked normal two weeks after stopping (15 people, maker-funded, so Moderate evidence). The intuition borrows from hormone systems that downregulate; it doesn't apply here.

Why light beats the pill. The clock is set primarily by light on the retina: morning bright light advances and stabilises the rhythm and triggers a >50% cortisol rise (daytime alertness, mood); evening light suppresses and delays your own melatonin. The behaviours fix the upstream clock and deliver mood, alertness, and metabolic co-benefits; the pill nudges one downstream signal.

Risks And Contraindications

• Children & adolescents — the strongest concrete harm signal. US paediatric accidental ingestion has surged (Lelak 2022: +530% poison-centre calls 2012–2021; CDC MMWR 2024: ~10,930 ED visits 2019–2022, gummies 47.3% of cases) — keep it locked away. Beyond overdose, there is no long-term safety data in developing children and an unresolved puberty/HPG-axis concern (melatonin restrains the reproductive axis; the nocturnal decline is a hypothesised puberty trigger — Boafo 2019 found zero clinical studies testing this in children). Legitimate paediatric use is sleep-onset insomnia in autism/ADHD under specialist supervision, not casual use in a healthy child.
• Glucose/metabolic (see Evidence): impaired glucose tolerance near food; extra caution for diabetics, pre-diabetics, and MTNR1B carriers. Take away from meals.
• Drug interactions: fluvoxamine (CYP1A2 inhibitor) raises melatonin up to ~17× (highest-risk); plus anticoagulants/warfarin (monitor INR), antihypertensives (nifedipine efficacy reduced), sedatives/benzodiazepines/Z-drugs (additive CNS depression; potentiates zolpidem impairment), diabetes medications (additive glucose effects), and possibly immunosuppressants (melatonin is immune-stimulating).
• Elderly: next-day sedation and fall risk at higher doses; daytime/driving impairment — keep dose low (1–3mg PR) and timing correct.
• Pregnancy/breastfeeding: data inadequate — not proven harmful, not established safe; breastfeeding essentially unstudied. Avoid routine use.
• Product quality: US content runs 74–347% of label, sometimes with CBD or (historically) serotonin — you often don't know your true dose.
• Milder/contested: vivid dreams/nightmares, headache, next-day dizziness (mild, dose-related); seizure-threshold effect is genuinely debated (caution in epilepsy, but not established as pro-convulsant); autoimmune/immune-stimulation is theoretical (Tier 4).
• This entry informs the decision; it is not medical advice. Chronic insomnia warrants behavioural therapy (CBT-I), not a standing supplement.

Controversy

Nature: two opposite overstatements, a commercial dosing distortion, and a regulatory split.

Position A — "Take 5–10mg every night to sleep." The mass-market take.
• Best evidence: melatonin modestly helps onset and isn't habit-forming.
• Where it's wrong: it's a chronobiotic, not a sedative; benefit is ~7 min and plateaus ~4mg; AASM advises against it for chronic insomnia; megadoses add grogginess and (with food) a glucose hit.

Position B — "Never use melatonin; it suppresses your own production and is addictive." The wellness-contrarian take.
• Best evidence: optimising endogenous production via light first is genuinely the right priority, and "is it worth it?" is a fair question (often no).
• Where it's wrong: the suppression/dependence mechanism is not supported (Lemoine). The valid case against routine use is small benefit + real downsides + better free alternatives, not gland shutdown.

Position C — "It's a harmless supplement." The reassuring take this entry used to lean toward.
• Best evidence: for a healthy adult, short-term, low-dose, correctly-timed, it is low-risk and non-habit-forming.
• Where it's wrong: that carve-out is narrow. It is not zero-downside for children, diabetics/MTNR1B carriers, people on interacting drugs, or pregnancy — and US product quality means you may not even know your dose. Most of the world regulates it as a medicine for a reason.

The funding/bias dimension: the US supplement industry profits from high-dose, frequently-mislabelled OTC products and the "harmless sleep aid" framing; the contrarian wellness market profits from the dependence scare and "melatonin-free" alternatives. The clean signal is mostly independent: the dose-response, glucose, product-content, and meta-analytic data, plus the non-US regulators who treat it as a hormone, with the maker-funded discontinuation study as support rather than independent evidence — converging on "small benefit, conditional risk, free levers first."

Realised Position: Melatonin is a clock signal, not a sleeping pill. The "it shuts down your own production / it's addictive" fear finds no support in the one long-term study that looked — and it is not a harmless everyone-supplement. For ordinary sleep trouble it isn't worth it: the benefit is ~7 minutes, guidelines advise against it for chronic insomnia, and the free circadian levers fix the cause and deliver mood/metabolic dividends a pill can't. Reserve it for genuine circadian problems (jet lag, shift work, delayed phase), at 0.3–0.5mg, timed early, away from food, quality-tested, short-term — with explicit exceptions for children, diabetics/MTNR1B carriers, interacting drugs, and pregnancy.

Cross-Pillar Connections

• Sleep / Circadian (circadian_rhythm_optimization, morning_sunlight_exposure): the upstream, free, multi-benefit levers that set the same clock — the recommended first move for ordinary sleep trouble.
• Sleep (cbt_i_program_overview): for chronic insomnia, CBT-I is first-line and melatonin is not recommended — this entry hands off there.
• Diet (light_exposure_vitamin_d): the parallel "sunlight does things a pill doesn't" theme; and the metabolic angle connects melatonin to glucose handling.
• Diet (meal_timing_consistency_and_circadian_alignment): the away-from-food rule and the late-eating × melatonin glucose interaction.
• Label literacy (supplement_form_elemental_dose_and_bioavailability): the melatonin-product mislabelling is the same "the label can mislead" principle, here a de-facto safety issue.

What would change our mind

Falsifiability: explicit upgrade/downgrade criteria from source

• We'd revise the "no suppression / no dependence" stance if long-term studies demonstrated durable endogenous downregulation or a withdrawal syndrome — current data (mainly Lemoine) show neither.
• We'd soften the glucose caveat if larger trials showed the post-meal glucose effect is clinically trivial at low chronobiotic doses taken away from food.
• We'd upgrade it toward "worth it" if trials showed melatonin meaningfully outperforming the free circadian levers for ordinary insomnia (current evidence and AASM say the opposite).
• We'd relax the children caution if long-term paediatric safety and the puberty question were resolved by RCTs (they are currently unstudied).
• What would NOT move us: the modest effect size and the product-quality/regulatory facts are well established. The no-suppression finding is Moderate and would move with a larger independent study.

Industry bias note

Structural incentives the evidence base may reflect

The profit pressure is at the megadose OTC end: 5–10mg gummies marketed as nightly sleep aids, frequently mislabelled (up to +347%), implying "more = better sleep." The contrarian end has its own market — "melatonin-free" products and supplement-vitamin sellers pushing the "your gland will shut down" fear. The single most telling independent signal is the regulatory split: the US treats melatonin as a dietary supplement, while the UK, Japan, and Australia treat it as a medicine, mostly on prescription, and EU countries set their own rules — a structural disagreement that should make any "harmless supplement" claim suspect. Realised anchors on the independent data (Rubio-Sastre/Garaulet glucose; Erland & Saxena and Cohen/JAMA product content; Ferracioli-Oda and AASM efficacy), reads the maker-funded Lemoine discontinuation study as supporting rather than independent evidence, and weighs these over both the megadose marketing and the dependence scare.

Melatonin: A Clock Signal Worth Skipping for Most People