melatonin · graded evidence · pulled 2026-08-31

Melatonin

The only drug in this catalog where the pharmacology argues for less, not more. The sleep dose-response curve is flat from 0.3 mg; timing predicts response almost perfectly; and the megadose tier — 20 mg and up — is oncology dosing with its own safety paper trail.

The short version
  • 0.1, 0.3, 1, 10 mg: all equally effective for sleep induction — the curve is flat from physiological doses on up. DATA
  • 0.3 mg reproduces normal nocturnal blood levels; 3 mg causes hypothermia and daytime spillover. DATA
  • Timing explains 94% of the variance in response (r²=0.94) — dose explains almost none. DATA
  • 20 mg/day is an oncology dose with a real trial record — and 300 mg/day was stopped early for liver-enzyme elevations in 3 of 4 patients. DATA

The flat curve

TrialDoses comparedFinding
Dollins 1994, PNAS0.1 / 0.3 / 1 / 10 mgall equally effective; 0.1 and 0.3 mg reproduce peak nocturnal plasma levels
Zhdanova 19950.3 vs 1 mgequivalent sleep promotion; no REM suppression, no hangover
Zhdanova 2001, elderly0.1 / 0.3 / 3 mg0.3 mg restored sleep efficiency (P<0.0001); 3 mg caused hypothermia + next-day spill-over
Mundey 2005, insomniacs0.3 vs 3 mgequivalent efficacy; response driven by dosing time, r²=0.94
Four independent placebo-controlled trials, one message: above the physiological dose the sleep curve does not rise. More milligrams buy longer plasma exposure — which is a spillover problem, not a potency gain.

How big is the effect at all

Meta-analysis: latency −7 min, total sleep +8 min DATA
19 RCTs, n=1,683 (Ferracioli 2013): sleep latency −7.06 min, total sleep time +8.25 min. Melatonin is a phase-setting hormone with a modest direct hypnotic effect — the people who benefit most are those whose internal clock is misaligned (delayed phase, shift, jet lag), which is exactly what the r²=0.94 timing result reflects.

Pharmacokinetics — why formulation changes the question

IR: t½ 1 hour. SR: t½ 5 hours, half the peak DATA
Direct comparison (Thanawala 2024): immediate-release half-life 1.01 h, Tmax 0.87 h; sustained-release half-life 5.10 h with Cmax roughly halved. A 10× dose taken as IR is a spike; the same taken as SR is a long exposure — spillover into the daylight window is a formulation-and-dose problem, and only the low doses avoid it by construction.

The megadose tier — what is actually known

ExposureDoseOutcome
Dollins 1993, daytime dosing10–80 mgimpaired, not improved performance — high daytime levels are a sedation-impairment state
Oncology RCT (Messina 2010)20 mg/dtumor-response benefit signal — the setting 20 mg actually comes from
Ninomiya 202630 mg/d × 49 dclean safety panel, n=12, short duration
MELATOMS-1, Bejarano 2026300 mg/dstopped after 8 enrollments — 3 of 4 melatonin patients developed hypertransaminasemia (reversible)
Between 0.3 and 300 mg the sleep benefit does not move; the exposure does. The 300 mg halt is the only hard toxicity signal on record, and it appeared at 1000× the physiological dose.

Interactions and endocrine notes

Cleared by CYP1A2 — check co-medication DATA
Melatonin is a CYP1A2 substrate; fluvoxamine inhibits its clearance with Ki ≈ 0.02 µM — an order-of-magnitude interaction that turns an ordinary dose into a long exposure. Any 1A2 inhibitor or SSRI co-medication changes the effective dose.
Growth-hormone interaction: real but pathway-specific DATA
Classic endocrinology (Valcavi): melatonin roughly doubles GHRH-stimulated GH release — via somatostatin-pathway modulation — while having no direct effect on pituitary GH secretion in vitro. Relevant to anyone stacking it with GH: the interaction is documented at physiological melatonin levels, not studied at 20 mg.
Reproductive axis: rodent suppression, human male adolescent null DATA
In female rodents, 1 mg/kg suppresses the HPG axis. In human male adolescents, no suppression signal was found. No tolerance/desensitization study with chronic dosing exists — receptor downregulation under months of supraphysiologic exposure remains uncharacterized.

What does not exist

No chronic high-dose trial · no desensitization data · no adolescent megadose safety study DATA
Verified by search (2026-08-31): no trial of melatonin above 10 mg/day running longer than 8 weeks, no receptor-desensitization study under chronic dosing of any strength, and no safety study of high-dose melatonin in adolescents. The 20 mg tier's chronic safety case rests on oncology cohorts of sick adults and one 49-day n=12 panel — plus the flat sleep curve that gives no compensating benefit for the exposure.