Melatonin Side Effects: A Dose Problem the Shelf Makes Worse

By Erin Rose · Published September 18, 2026 · Reviewed against primary sources · Methodology · About Us

Informational summary of published trials, systematic reviews, case series and government surveillance data, not medical advice. If a child has swallowed melatonin, call Poison Control at 1-800-222-1222 now. If you take an anticoagulant, an immunosuppressant, a seizure medicine or blood-pressure medicine, ask your prescriber before you start melatonin rather than after.

Choosing a dose rather than troubleshooting one? The melatonin dosage guide owns the amount and the timing, including why the hour you take it matters as much as the milligrams →

Common effects, and the dose they appear at

Melatonin is a hormone your pineal gland already makes, and a supplement is a second, larger dose of it arriving at a time you choose. That is why its side-effect list is short, mild and almost entirely about quantity and timing. The most careful count comes from a systematic review of 37 randomized, placebo-controlled trials of melatonin for sleep disorders, using doses from 0.15 to 12 mg and following people for up to 29 weeks (Besag 2019, PMID: 31722088).

The grogginess deserves its own explanation, because it is the complaint that brings most people to a page like this. Melatonin’s job is to mark the biological night; the effect ends when the level falls. In a trial of 30 adults over 50, the physiologic 0.3 mg dose restored sleep efficiency and brought overnight melatonin back into the normal night range, while the pharmacologic 3 mg dose also improved sleep but induced hypothermia and left plasma melatonin elevated into the daylight hours (Zhdanova 2001). A second study put a number on the daytime version: 15 reserve pilots given 3 mg before a short nap had mildly impaired psychomotor performance and felt significantly sleepier for 2 to 4 hours afterwards (Nave 2002, PMID: 11844575). A morning hangover is not an idiosyncratic reaction. It is the dose still working.

The other systematic review worth knowing is the one that looked specifically at whether melatonin does anything bad on purpose. Foley and Steel screened 50 controlled studies that reported a statistical analysis of adverse events: 26 found no statistically significant adverse event at all, and 24 found at least one, most often relating to fatigue, mood, or psychomotor and neurocognitive performance (Foley 2019, PMID: 30670284). The endocrine and cardiovascular signals they did find were tied to dose, to the time of day it was taken, and to interaction with blood-pressure medicine. That is the shape of melatonin’s risk: small, reversible, and governed by how much and when.

Serious effects, and who is actually at risk

The honest summary for a healthy adult is that the serious-harm evidence is thin. The meta-analysis of high-dose trials pooled 79 studies and 3,861 participants taking 10 mg or more; only 4 met its low-risk-of-bias bar, and in those, melatonin showed no detectable increase in serious adverse events (rate ratio 0.88, 0.52–1.5) while the ordinary adverse events rose to 1.4 times placebo (1.15–1.69) (Menczel Schrire 2022). That is a good safety signal with a large caveat attached by the authors themselves: 29 of the 79 studies (37%) never said whether adverse events happened at all. Absence of reporting is not evidence of absence.

Where melatonin does cause documented harm, it is almost always in a specific group.

• Autoimmune disease, particularly myasthenia gravis. A referral centre described 3 patients whose myasthenia gravis worsened within days to weeks of starting melatonin; two stopped it without improving over the following week, and one needed intravenous immunoglobulin (Nedkova-Hristova 2020, PMID: 33148186). This is a case series from one centre, not a rate, and it cannot tell you how often this happens. It is reason enough to ask first if you have an autoimmune diagnosis or take an immunosuppressant.
• Epilepsy and anticoagulation. The NIH National Center for Complementary and Integrative Health states that people with epilepsy and people taking blood-thinning medication need to be under medical supervision when taking melatonin (NCCIH). We could not find a controlled human trial quantifying either risk, and we are not going to invent a number for one.
• Pregnancy and breastfeeding. A scoping review found 15 human studies in total (8 in pregnancy, 7 in breastfeeding), none with insomnia as a primary outcome, and concluded that use is probably safe while calling for actual trials (Vine 2022, PMID: 34730672). The earlier safety review reached the opposite practical conclusion from the same absence of data and advised against use (Andersen 2016). With about 4% of pregnant people already using it, the honest statement is that nobody has tested this properly; decide it with your obstetrician, not with a search result.
• Long-term nightly use. The longest controlled trial followed 791 adults on 2 mg prolonged-release melatonin for 6 months and found no clinically relevant safety difference from placebo and no loss of effect over time (Wade 2010, PMID: 20712869). Beyond six months, and beyond that formulation and dose, there is no controlled evidence in either direction.

One more thing belongs here because it is a reason to stop rather than a side effect. The American Academy of Sleep Medicine’s guideline on drugs for chronic insomnia reviewed melatonin and suggests clinicians not use it for sleep-onset or sleep-maintenance insomnia, a weak recommendation reflecting weak evidence of benefit (Sateia 2017, PMID: 27998379). If you have been taking melatonin nightly for months for ordinary insomnia and getting side effects from it, the trade you are making is worse than it looks.

The dose you took may not be the dose you chose

Every statement above ties an effect to a number of milligrams. That chain breaks if the pill does not contain what the panel says. Two independent analyses tested it.

The first assayed 31 commercial melatonin supplements. Content ranged from 83% below to 478% above the labeled amount; more than 71% missed the label by more than a 10% margin; lot-to-lot variation within a single product reached 465% (Erland 2017, PMID: 27855744). The second tested 25 melatonin gummies bought in the United States: 22 of them (88%) were inaccurately labeled, measured content ran from 74% to 347% of what the label declared, and 1 contained no detectable melatonin (Cohen 2023, PMID: 37097362).

Read that against the side-effect evidence and it becomes a dosing problem rather than a labeling one. If a 5 mg gummy delivers 347% of its label, the person taking it is in the 10 mg-plus territory where the adverse-event rate ran 1.4 times placebo, without having chosen that. The practical response is not to chase a milligram figure but to buy a product whose content someone else has checked: a USP Verified or NSF mark is the only routine third-party confirmation available on this shelf. Our dosage guide covers the assay evidence in full; what matters here is that an unverified label and a side effect are the same problem seen from two ends.

Interactions that have human data behind them

Melatonin is cleared mainly by the liver enzyme CYP1A2, so anything that blocks that enzyme raises the dose you effectively took. Below are the interactions with measured human pharmacology or a controlled trial, and only those. Lists elsewhere are longer; most of the extra entries are theoretical.

Alcohol is not in the table because we found no controlled human study of the combination worth quoting. It is reasonable to expect two sedating agents to add up, and alcohol suppresses your own melatonin release, but that is mechanism, not measurement.

We read 819 labels: the shelf sells well above the dose the trials used

Original research Every on-market melatonin supplement label in the NIH Dietary Supplement Label Database whose product name contains the term (1,804 on the market, 819 read, 808 with a parseable melatonin quantity, 768 with a daily serving direction, 241 brands), dose taken from the Supplement Facts panel, compared against the 0.5–3 mg range used in trials.

Every guide, ours included, tells a new user to start at about 1 mg. Of the 808 labels that declare a dose, 123 sell that dose or less and 685 sell more. Going further up, 370 of 808 sell more than 3 mg, the top of the range the trials used at all. The median label sells 3 mg per serving and 3 mg a day as directed; the 90th percentile sells 10 mg; the largest single label we read sells 60 mg, 60 times the typical studied dose and 15 times the point where the sleep benefit stopped improving.

Two consequences follow for someone with a bottle in hand. The first is that the dose most guides recommend is genuinely hard to buy: 123 labels out of 808 is the whole low-dose shelf, so “just take 1 mg” often means splitting a tablet. The second is about gummies: 87 of the labels we read are gummies or jellies, a descriptive count of the shelf rather than a validated finding. The JAMA analysis of gummies found the measured amount often differed from the label.

One thing this read cannot tell you, and one thing it will not. It cannot tell you what is in the capsule — that is the assay question above, and it is a different number. And one of the three findings we tested, that the median dose as directed exceeds the median per serving, did not hold in both halves of the sample, so no figure for it appears anywhere on this page.

Method: 819 labels retrieved from the NIH Dietary Supplement Label Database (DSLD) v9 on 2026-08-28 for the term “melatonin”; 808 declare melatonin with a parseable quantity and 768 state a daily serving direction. Every share is stated against the 808 labels that declared a dose, never the 819 read. What it cannot tell you: which products sell most, what the capsule actually contains, or anything about products not filed with the NIH. Data released under CC BY 4.0.

What to do, matched to what you have

There is no withdrawal syndrome to manage if you decide to stop. The six-month trial found no rebound insomnia and no withdrawal effects when prolonged-release melatonin was discontinued, and the same was true in the 170-patient trial that preceded it (Lemoine 2007, PMID: 18036082). You can simply stop.

The product links below go to Amazon and we may earn a commission if you buy. It never changes which product we pick or what we say about it. How we choose.

Frequently asked questions

Related guides

• Melatonin dosage guide: how much, when to take it, and the full label-accuracy evidence
• Best melatonin supplements: every verified pick ranked by cost per night
• Magnesium for sleep: the alternative when melatonin is the wrong tool
• The anxiety and sleep stack: what combines and what does not
• Magnesium side effects and vitamin D side effects: the same treatment for the other two clusters
• All melatonin guides

Sources