Sleep is where these earn their place — replicated, modest improvements in how fast you fall asleep and how rested you wake. Mood and next-day energy come along for the ride; nothing transformative, but the dose, timing, and choice of substance matter more than most users realize. The catch: the loudest options on the shelf — 10 mg melatonin gummies, apigenin marketing — are not the ones with the cleanest evidence underneath.
Five different routes to the same outcome, which is why people stack them. Magnesium turns down the excitatory tone in your nervous system — the same brake your body uses to ease itself toward sleep at night. L-theanine, an amino acid from green tea, shifts your brain into a calmer kind of wakefulness; the EEG signature looks like ten minutes into a good meditation Hidese 2019. Glycine drops your core body temperature by a fraction of a degree, the same shift a warm bath produces an hour before bed Kawai 2015. Apigenin, a flavonoid from chamomile and parsley, binds the same brain site that valium binds — only weakly Salehi 2019. And melatonin is the body's darkness signal: not a sleeping pill in the way most people use it, but a chemical message that says it's bedtime, the lights are out, you can let go.
Each of these mechanisms hits a different part of the falling-asleep problem. Magnesium and apigenin lower the noise in your brain. Glycine cools you down. L-theanine takes the edge off the day. Melatonin tells your body what time it is. None of them is sedation in the prescription-sleeping-pill sense — none of them knocks you out the way an over-the-counter antihistamine "PM" aid does, at the cost of the anticholinergic burden those quietly carry — and that's both why their side-effect profiles are good and why their effects feel small.
What the trials actually show
None of these produce dramatic numbers. They produce small, replicated ones.
Magnesium has one solid trial: forty-six older adults with insomnia, eight weeks of 500 mg a day, falling asleep about seventeen minutes faster than placebo, with better sleep efficiency and lower next-morning cortisol Abbasi 2012. A 2021 systematic review pulled three such trials together and confirmed the direction of effect, while honestly rating the certainty of the evidence as "low" — small samples, mixed magnesium forms, room for bias Mah & Pitre 2021. The effect is biggest when the user's dietary magnesium was probably low to start; in fully repleted people, the signal fades toward placebo.
L-theanine has a clean placebo-controlled crossover trial: 200 mg a day for four weeks, healthy adults under stress, better Pittsburgh Sleep Quality scores and lower subjective stress Hidese 2019. A 2015 review put L-theanine among the strongest natural-sleep candidates after melatonin on the basis of mechanism plus safety profile Rao 2015.
Glycine has three small Japanese trials, all positive, all from the same research group: 3 g before bed, faster sleep onset on polysomnography, more slow-wave sleep early in the night, less next-day fatigue — even after a partial-sleep-restriction night Yamadera 2007, Inagawa 2006, Bannai 2012. Coherent mechanism, thin replication, and the manufacturer of the studied glycine was an author affiliation. Defensible but not robust.
Apigenin is the exception. There are no human sleep trials of purified apigenin. The chamomile-extract evidence is real but rides on multiple active compounds; the longest trial was eight months of high-dose chamomile for generalised anxiety disorder, where it reduced symptom severity but didn't prevent relapse Mao 2016. The popular 50 mg pre-bed apigenin dose is extrapolated from rodent data and from podcast endorsements, not from a controlled human sleep study.
How to take them well
Every supplement on this list has a "taken seriously" dose and a "thrown into a gummy" dose. The taken-seriously doses below are what the trials actually used.
For a first pass, magnesium glycinate plus 0.3–0.5 mg melatonin (if your sleep timing is off) covers most of what these supplements can do for most users. Adding glycine or L-theanine is reasonable if the first pair isn't enough; layering apigenin is the most speculative move on the list.
Three things to unlearn
Melatonin is not a sleeping pill. It's a chemical message that tells your body it's nighttime. Take it at the right time in a dark room and it nudges you toward sleep; take it in a bright kitchen scrolling your phone and it does almost nothing, because the eyes are still telling your brain it's daytime. People who say "melatonin doesn't work for me" are usually using it as a sedative — which it isn't — and at five to thirty times the dose that the research is for AASM 2017.
More is not better. A 0.3 mg dose of melatonin restores your blood level to about what your body produces on a normal night. A 10 mg gummy puts you ten to a hundred times above that, and the body's receptors respond by becoming less sensitive — so the next night you need more, and the effect fades anyway Zhdanova 2001. Magnesium past about 400 mg starts producing diarrhea before it produces more sleep. Glycine and L-theanine haven't been tested above the studied doses, and there's no reason to expect a bigger pill to do bigger work.
The label dose isn't always the dose. When researchers measured the actual melatonin content of 31 commercial supplements, the amounts ranged from 83% below to 478% above what the label claimed; one in four also contained measurable serotonin, an unlisted controlled compound Erland & Saxena 2017. A 2023 follow-up tested melatonin gummies sold to children: 22 of 25 were off by at least 10%, and one gummy contained three and a half times the labelled dose Cohen 2023. US supplements aren't pre-approved for content; the FDA only acts after the fact. Stick to brands that publish independent test results — USP-verified, NSF-certified, or third-party-tested through ConsumerLab or LabDoor.
Who should skip parts of this
Outside these groups, the safety profile of magnesium, L-theanine, and glycine at the studied doses is among the best in the supplement category. Low-dose melatonin (0.3–0.5 mg) is also well tolerated; the safety concerns scale with the dose.
Cost and brand selection
This is one of the cheapest categories in the catalogue. A three-month supply of magnesium glycinate runs about $15–30; L-theanine about the same; melatonin $10–20 for six months or more at low dose; glycine powder $15–25 for two or three months at 3 g a day. Apigenin is the most expensive — $25–40 for a month or two. A full stack stays under $200 a year; a single substance under $50.
The decisive practical choice is the brand, not the substance. Because US supplements aren't tested for content before they ship, your magnesium-glycinate capsule could contain anywhere from a fraction of the labelled dose to several times it. The shortcut: look for a USP-verified or NSF-certified seal on the bottle, or check the brand against an independent tester (ConsumerLab and LabDoor publish periodic accuracy reviews). A slightly pricier verified brand beats a cheaper unverified one — most of what you're paying for is the assurance that the pill matches the label.
One regional note: outside the United States, melatonin is prescription-only in the EU, the UK, Japan, and Australia (Canada permits OTC sale). If you're travelling, don't expect to pick up a bottle at a foreign pharmacy.
Why it didn't work for you
The most common reason a sleep supplement "doesn't work" is that the thing keeping you awake isn't the thing the supplement fixes. Magnesium can't undo a 9 p.m. espresso. Melatonin can't outvote a bedroom full of overhead light. Glycine doesn't compete with the second half of a bottle of wine — alcohol disrupts the second half of the night more than any supplement compensates for. If you're scrolling in bed under a 6500 K LED until your eyes burn, the supplement is doing the hard work and the lights are doing the harder one.
The second most common reason is wrong substance for the problem. If you fall asleep fine but wake at 3 a.m. and can't get back, low-dose melatonin and L-theanine are weak picks — magnesium and a prolonged-release melatonin formulation have more to offer. If you fall asleep wired from the day, L-theanine is the better lead than magnesium. If your sleep timing has drifted late, melatonin is right but the dose is 0.3–0.5 mg taken five hours before bed, not 5 mg taken at the door of sleep.
The third reason is wrong dose. A 10 mg melatonin gummy may leave you groggy in the morning without giving you a better night; the lower dose is what the trials are about Zhdanova 2001. Magnesium oxide at the same milligram count as glycinate gets you a tenth of the magnesium. Apigenin at the marketed dose is probably below what would do anything in humans at all.
The fourth, less obvious reason: the placebo arm in good sleep trials gets a real effect too, and yours might be running quietly. Give a new supplement two weeks of consistent use before judging it; if there's nothing by then, switch or stop.
The bigger lever, named honestly
If you have actual insomnia — most nights, for weeks, with daytime consequence — supplements aren't the answer. The intervention that the American Academy of Sleep Medicine and the American College of Physicians both name as first line, with stronger evidence than any pill, is cognitive behavioural therapy for insomnia (CBT-I) Edinger 2021, Qaseem 2016. Six to eight weeks of a structured protocol; effect sizes that comfortably exceed any supplement on this page; results that persist after you stop. In-person delivery runs into the thousands; digital versions (Sleepio, Somryst, the free CBT-i Coach app) cost under $200 or nothing.
Below clinical insomnia, the sleep-hygiene basics outweigh supplements for most users: a cool dark bedroom, a consistent wake time, morning sunlight on the eyes within an hour of waking, evening light kept dim, caffeine cutoff by early afternoon, alcohol kept moderate and well before bed. These cost nothing and move bigger numbers than any pill in this entry. The honest framing of sleep supplements is as the small lever you reach for after the big ones are in place, not as a substitute for them.
If supplements and hygiene both fall short, the prescription rung is real: dual orexin receptor antagonists (suvorexant, lemborexant) have less rebound than older drugs; low-dose doxepin handles sleep-maintenance insomnia; trazodone is widely used off-label. All require a clinician, and none of them belong in a self-directed plan.
What changes if it works
The honest version of the payoff isn't dramatic. If you've matched the substance to your problem and got the dose right, here's what the first two weeks look like for most users.
The first night: probably nothing you can clearly attribute to the supplement. The placebo effect is real and quiet; so is the supplement effect; they're hard to tell apart on a single night. Don't read too much into either direction.
The first week: a small change in the shape of your evenings. You notice yourself drifting toward bed instead of fighting your phone, or your mind goes quieter at lights-out instead of replaying the day. You're not knocked out — that's not what these do — but the gap between "I should sleep" and "I'm asleep" closes by something like seven to fifteen minutes Ferracioli-Oda 2013, Mah & Pitre 2021. The mornings feel a fraction less hung-over from the night, especially if the supplement displaced a half-glass of wine or a late screen session.
By the second week: the social mirror starts to register. Your partner stops asking why you were up until 1 a.m. on a Tuesday. Your morning meeting goes a little easier. You stop reaching for the third coffee. None of this is a transformation. It's the version of you that already exists getting back twenty minutes of sleep and the next-day energy that comes with it.
The cases where the effect is bigger than this: shift workers and jet-lagged travellers using low-dose melatonin for what it's actually good at (resetting the clock), older adults with declining endogenous melatonin using prolonged-release at bedtime Lemoine 2007, and stressed adults whose pre-sleep anxiety was the bottleneck and who benefit from L-theanine taking that edge off Hidese 2019. The cases where the effect is smaller than this: anyone whose sleep problem has a structural cause the supplement can't reach — apnea, alcohol, anxiety as a clinical condition, a bedroom that fundamentally isn't dark or quiet.
Related entries worth a look once you've got the supplement question sorted: sleep hygiene fundamentals and morning light exposure are where most users get the bigger sleep gains; CBT-I is the first line for actual insomnia; the sleep apnea entry covers the most common hidden cause of "I sleep eight hours and still wake up tired"; alcohol and sleep covers why a nightcap costs you more than it gives. Specific stimulant-side topics — caffeine timing and afternoon caffeine cutoffs — sit upstream of most of these. None of this is supplement-shaped, and that's the point.
- — Many OTC sleep aids are sedating antihistamines that quietly raise anticholinergic burden.
- — Melatonin is the most misunderstood one here — a clock signal, not a sedative, and the gummy dose is wildly too high.
- — Since sleep-supplement labels frequently misstate the dose, look for a third-party seal before you trust one.
- — Ashwagandha overlaps with this shelf — sold for sleep and stress, with a similarly modest evidence base.
- — Magnesium is one of the five shelf staples here — modest, real, and cheaper bought on its own.
- — L-theanine shows up on both lists — here for winding down, there for calm focus. Same modest-but-real evidence, just a different time of day.
- — These buy 7 to 17 extra minutes — useful at the margins, but no supplement closes a chronic sleep gap. Fix the hours first.
Substance and claimed effects
This entry covers five over-the-counter sleep supplements that dominate the consumer market: magnesium (most often as glycinate, citrate, or threonate), L-theanine (an amino acid from green tea), low-dose melatonin (0.3–1 mg, the dose that approximates physiologic nocturnal peaks), glycine (an inhibitory amino acid taken 3 g pre-sleep), and apigenin (a flavonoid from chamomile and parsley). Claims across these supplements span four addressable outcomes: reduced sleep-onset latency, improved sleep continuity (fewer arousals, more slow-wave sleep), reduced next-morning grogginess relative to prescription hypnotics, and acceptable long-term tolerability. Each substance hits a different sub-mechanism — NMDA / GABA tone (magnesium, glycine, apigenin), alpha-wave-mediated relaxation (L-theanine), or the circadian phase-marker pathway (melatonin) — so the consequences this entry covers holistically are sleep, mood / stress, knock-on energy and focus the day after, plus the structural burden (cost, daily compliance) and evidence/controversy picture. Longevity and appearance dimensions are touched only indirectly — via sleep itself, the upstream driver — and are not where these substances earn their place. Two cross-cutting features matter throughout: melatonin is mechanistically a chronobiotic, not a sedative, and is misused as the latter; and the OTC supplement market in the US has documented quality-control failures that determine whether a labelled dose maps to what enters the body.
Evidence by addressing question
Mechanism
Magnesium. Magnesium is a cofactor for >300 enzymatic reactions and acts as a voltage-dependent NMDA receptor antagonist; it also potentiates GABA-A receptor binding and modulates melatonin synthesis via N-acetyltransferase activity in the pineal gland. The plausibility chain for sleep is: lower NMDA-mediated excitatory tone + higher inhibitory tone at GABA-A → easier transition to sleep onset. Mechanism strongest in true magnesium deficiency (RDA ~310–420 mg/day; surveys suggest ~50% of US adults consume below the EAR, though frank deficiency is rare). The glycinate, citrate, and threonate salts are better absorbed than oxide; threonate uniquely crosses the blood-brain barrier in animal models. Effect-size mechanism is modest — magnesium is not a sedative, it is a sleep-architecture facilitator Rao 2015, Boyle 2016.
L-theanine. Structural analogue of glutamate and glutamine. Crosses the BBB; binds glutamate receptors weakly and modulates GABA, dopamine, and serotonin release. The signature EEG signature is a clear increase in alpha-wave power within 30–45 minutes of a 200 mg oral dose — the brain state associated with "wakeful relaxation," what users describe as "the edge coming off." Not directly sedating: theanine does not increase delta or slow-wave activity in awake EEG. The proposed sleep route is indirect: lower pre-sleep autonomic arousal → easier sleep onset, fewer middle-of-night arousals driven by sympathetic tone Hidese 2019, Rao 2015.
Low-dose melatonin. Endogenous melatonin is secreted by the pineal gland on a circadian schedule, peaking 2–4 a.m. at plasma concentrations of ~50–200 pg/mL. Exogenous melatonin acts on MT1 (sleep promotion) and MT2 (circadian phase shift) receptors in the suprachiasmatic nucleus. The critical dose-response point: 0.3 mg restores plasma melatonin to roughly endogenous nocturnal levels; 3–5 mg gummies produce supraphysiologic levels 10–100× normal, with the excess clearing slowly because melatonin has a half-life of ~40 min but receptor desensitisation at high doses paradoxically reduces sleep-promoting effect. The chronobiotic effect — phase-shifting the circadian clock — is dose-independent above ~0.3 mg Zhdanova 2001, Burgess 2010. Melatonin is functionally a "darkness signal," not a sedative; this is the central misconception in consumer use.
Glycine. Inhibitory neurotransmitter, NMDA receptor co-agonist (paradoxically — at the glycine binding site on NR1 it modulates rather than activates). Animal work shows oral glycine reaches the brain and acts on NMDA receptors in the suprachiasmatic nucleus, producing peripheral vasodilation → distal skin temperature rise → core body temperature drop. Falling core temperature is a powerful sleep-onset signal in humans (the same mechanism behind the "warm bath 1–2 hours before bed" recommendation). The 3 g pre-bed dose was chosen in the original Japanese studies to maximise this thermoregulatory effect Kawai 2015, Bannai 2012.
Apigenin. Flavonoid; partial agonist at the benzodiazepine binding site of the GABA-A receptor — same site, lower efficacy than diazepam. In vitro and rodent data show anxiolytic effects without the motor-impairment signature of full benzodiazepine agonists. Concentrations needed for the GABA effect in humans are speculative: a 50 mg oral apigenin dose has been measured to produce plasma concentrations in the low nanomolar range, two to three orders of magnitude below the concentrations that produced behavioural effects in rodent models. Apigenin also has documented in vitro estrogenic activity, CYP1A2 / CYP2C9 / CYP3A4 inhibition, and aromatase inhibition — none of these have been characterised at human supplement doses, but they complicate the "harmless flavonoid" framing Salehi 2019.
Evidence
Magnesium. The flagship trial is Abbasi 2012: 46 elderly subjects with primary insomnia, 500 mg/day magnesium oxide vs placebo for 8 weeks. Significant improvements in ISI (Insomnia Severity Index), sleep efficiency, sleep time, and early-morning awakening; serum renin and melatonin rose, serum cortisol fell. The 2021 systematic review and meta-analysis pooled three RCTs (N=151 total) and found a 17-minute reduction in sleep-onset latency favouring magnesium, with the authors rating the certainty of evidence as "low" by GRADE due to small sample sizes, heterogeneous formulations, and high risk of bias Mah & Pitre 2021. Effects are stronger when baseline magnesium intake is low; little benefit signal in repleted individuals. The Boyle 2016 systematic review separately documented magnesium's anxiolytic effect, which probably contributes to the sleep effect via reduced pre-sleep autonomic arousal Boyle 2016, Abbasi 2012.
L-theanine. Hidese 2019 ran the cleanest RCT to date: 30 healthy adults, 200 mg/day for 4 weeks, double-blind crossover. PSQI (Pittsburgh Sleep Quality Index) global score improved significantly, driven by sleep latency and sleep disturbance subscales; subjective stress also dropped. Earlier work by Kim et al. and Lyon et al. in pediatric ADHD samples showed similar sleep-quality improvements. Rao 2015 reviewed the broader literature and called L-theanine the strongest natural-sleep-aid candidate after melatonin on the basis of safety profile + acute anxiolytic effect. Evidence is consistent in direction (small-to-moderate improvement in subjective sleep quality) but the trials are uniformly small and short, and PSQI-based outcomes are subject to expectancy bias Hidese 2019, Rao 2015.
Low-dose melatonin. The largest meta-analyses (Ferracioli-Oda 2013, n=1683 across 19 RCTs; Brzezinski 2005; Auld 2017) converge on the same headline numbers: melatonin reduces sleep-onset latency by ~7 minutes, increases total sleep time by ~8 minutes, and improves subjective sleep quality. Effect sizes are small but consistent and reproducible Ferracioli-Oda 2013, Brzezinski 2005, Auld 2017. The strongest evidence is for circadian disorders (delayed sleep phase, jet lag, shift work) rather than primary insomnia. The AASM 2017 clinical practice guideline explicitly recommended against melatonin for sleep-onset or sleep-maintenance insomnia in adults (weak recommendation against, very low quality of evidence) — a stance that has aged poorly with the OTC boom but reflects the actual trial evidence at the time AASM 2017. Prolonged-release 2 mg melatonin (Circadin, EMA-approved 2007 for ≥55) showed durable sleep-quality improvement and morning-alertness benefit in Lemoine 2007 — a population whose endogenous melatonin secretion is declining Lemoine 2007.
Glycine. Three small Japanese trials anchor the literature. Yamadera 2007 (N=11, crossover): 3 g glycine 1 h before bed shortened PSG-measured sleep onset, increased slow-wave sleep early in the night, and reduced next-day sleepiness. Inagawa 2006 (N=15) replicated the subjective effects (better sleep quality, less daytime fatigue) by questionnaire. Bannai 2012 reviewed the mechanism and added a workplace cognitive-performance arm: subjects given 3 g glycine after a partial-sleep-restriction night showed less self-reported fatigue and better PVT performance the following day Yamadera 2007, Inagawa 2006, Bannai 2012. Three small trials from one research group is not a robust evidence base, but the mechanism (core-temperature drop) is independently established and there have been no contradictory trials. No large independent replication exists.
Apigenin. Direct human evidence for apigenin as a sleep aid is essentially nil — no RCTs of pure apigenin for primary sleep outcomes. The evidence rides on chamomile, in which apigenin is one of several active compounds. Mao 2016 ran a 38-week RCT of long-term chamomile extract (1500 mg/day standardised) for generalised anxiety disorder; it did not prevent relapse vs placebo but reduced symptom severity. Earlier shorter trials (Amsterdam 2009, 2012) showed acute anxiolytic effects of chamomile in GAD. The leap from "chamomile extract reduces anxiety symptoms" to "apigenin at 50 mg improves sleep" is a leap, not a finding Mao 2016, Salehi 2019. The apigenin-specific dosing popularised in the sleep-supplement space (50 mg pre-bed) derives from extrapolated rodent data and the inferences of a small number of high-profile podcast hosts, not from controlled human sleep trials.
Protocol
Magnesium: 200–400 mg elemental magnesium, 30–60 min before bed. Glycinate (best tolerated, fewer GI effects), citrate (more laxative effect — useful if constipated, problematic otherwise), or threonate (more expensive, BBB-crossing in animal data). Avoid oxide form — ~4% bioavailability vs ~40% for organic salts. Tolerable upper intake from supplements (per IOM) is 350 mg/day to avoid diarrhea; the 500 mg/day doses in trials like Abbasi 2012 are above this and produced GI effects in some subjects.
L-theanine: 100–400 mg, 30–60 min before bed; can also be taken alongside afternoon caffeine to blunt jitter. No tolerance documented. Pharmacokinetic peak ~50 min.
Low-dose melatonin: 0.3–0.5 mg is the physiologic-restoration dose for sleep onset, taken 30–60 min before desired sleep. For circadian phase-shifting (jet lag, delayed sleep phase), 0.5 mg taken ~5 hours before habitual sleep onset advances the clock; the exact timing matters more than the dose. The popular 3–10 mg gummies are 6–30× the dose needed and more likely to produce next-day grogginess. Prolonged-release 2 mg is the formulation studied in older adults Zhdanova 2001, Burgess 2010, Lemoine 2007.
Glycine: 3 g (one heaped teaspoon of powder, or capsules), 30–60 min before bed. Sweet taste — palatable in water. The 3 g dose comes from the original Japanese trials; no clear dose-response data above or below this.
Apigenin: Marketed at 50 mg pre-bed; no validated human dose. Chamomile tea or extract (300–1500 mg standardised) has more direct human evidence than purified apigenin.
Contraindications
Magnesium: Reduced renal clearance is the main contraindication — moderate-to-severe CKD risks hypermagnesemia. Concomitant use with potassium-sparing diuretics or bisphosphonates affects absorption / interactions. Magnesium binds tetracycline / quinolone antibiotics — separate dosing by 2 hours.
L-theanine: No major contraindications established. Theoretical risk of additive hypotensive effect with antihypertensives. Pregnancy data sparse.
Melatonin: Pediatric exposures rose 530% from 2012–2021 in US poison-centre data Lelak 2022, including five deaths; gummies are the dominant form. Caution in autoimmune disease (theoretical immunostimulant effect), pregnancy, anticoagulants (warfarin interaction documented). Daytime drowsiness is dose-dependent. Suppresses LH/FSH at high doses (concern for adolescents). Banned for OTC sale in the EU, UK, Japan, Australia — prescription-only in most developed markets.
Glycine: Minimal contraindications. Theoretical caution with clozapine (glycine antagonises clozapine's effect in trials of schizophrenia adjunct). Sweet taste affects diabetic carbohydrate awareness but glycine itself is not glycemic.
Apigenin: Estrogenic activity in vitro — concern in hormone-sensitive conditions (breast cancer, endometriosis) is theoretical but not characterised in humans. CYP enzyme inhibition could affect drug clearance — relevant for narrow-therapeutic-window drugs (warfarin, certain antiepileptics). Less characterised safety profile than the others.
Misconceptions
"Melatonin is a sleeping pill." Mechanistically false. Melatonin is the body's darkness signal — it shifts the circadian phase and weakly promotes sleep onset within the existing biological window. It does not produce sedation in the benzodiazepine / Z-drug sense. People who report "melatonin doesn't work for me" are usually trying to use it as a sedative, against the actual mechanism. Misconception is reinforced by the OTC presentation (gummies, sleep-aid branding) and by the gummy-dose escalation (10 mg, 12 mg) that mimics sedative-style dosing AASM 2017.
"More is better." For melatonin this is actively false — supraphysiologic doses give worse outcomes than 0.3 mg in head-to-head trials Zhdanova 2001. For magnesium past ~400 mg, the marginal effect on sleep flattens and the diarrhea risk rises sharply. For glycine and L-theanine, no clean dose-response data exists above the studied doses.
"Natural means safe." Apigenin is the cleanest counterexample — estrogenic, CYP-inhibitory, and used in supplement form with essentially no safety surveillance. Melatonin's pediatric ingestion data is the public-health counterexample at scale Lelak 2022.
"The label dose is the dose." Erland & Saxena 2017 measured actual melatonin content across 31 commercial supplements: contents ranged from –83% to +478% of labelled dose; one in four also contained measurable serotonin, an unlisted controlled compound in many jurisdictions Erland & Saxena 2017. Cohen 2023 repeated the measurement on melatonin gummies: 22 of 25 contained ≥10% deviation from the label, one gummy contained 347% of the labelled amount Cohen 2023. The US supplement market does not pre-approve content; the FDA acts post-market.
"Supplements solve insomnia." The first-line evidence-based treatment for chronic insomnia is cognitive behavioural therapy for insomnia (CBT-I), endorsed by both the American College of Physicians and the AASM with stronger evidence than any pharmacologic option Qaseem 2016, Edinger 2021. Supplements are a useful adjunct for sleep-hygiene-resistant occasional poor sleep, not a substitute for behavioural treatment of clinical insomnia.
Failure modes
The dominant failure mode across all five is using the supplement to paper over a sleep problem whose root cause is behavioural (irregular schedule, late caffeine, evening light exposure, alcohol use), environmental (warm/bright bedroom, partner snoring, kids waking), or clinical (untreated sleep apnea, anxiety disorder, restless legs, hyperthyroidism, late-stage perimenopause). Adding magnesium to a 1 a.m. alcohol-impaired sleep doesn't fix the alcohol; adding melatonin to a brightly-lit phone-scrolling routine doesn't fix the light input the SCN is reading as daytime.
Substance-specific failure modes: (1) Melatonin taken at the wrong time can worsen circadian alignment — taken in the morning, it delays the clock; taken too early before sleep, the half-life means the phase-shifting effect lands during waking hours Burgess 2010. (2) Magnesium oxide is bought because it is cheapest but is the worst-absorbed form. (3) Glycine compliance fails for people who don't like sweet taste or remember to dose 1 h before bed. (4) Apigenin "didn't work" likely because the dose is sub-therapeutic. (5) L-theanine "didn't work" often means the user expected sedation — the effect is anxiolytic, not soporific.
Alternatives
The first-line alternative to all of these is CBT-I — cognitive behavioural therapy for insomnia, the only intervention rated "strong recommendation, moderate to high quality of evidence" by the AASM for chronic insomnia Edinger 2021. Digital CBT-I (Sleepio, Somryst, CBT-i Coach) brings cost from ~$2000 in-person to <$200. Effect sizes substantially exceed any supplement.
Sleep hygiene interventions (cool dark bedroom, consistent wake time, morning light, evening light reduction, late caffeine cutoff, alcohol limit) carry larger effect sizes than supplements for most users — the supplement category competes with hygiene improvements that cost nothing.
Prescription options when supplements fail: dual orexin receptor antagonists (suvorexant, lemborexant — newer mechanism, less rebound), low-dose doxepin (3–6 mg, sleep maintenance), trazodone (off-label, common), Z-drugs (zolpidem etc — short courses only, AASM warns against chronic use). All require clinician evaluation.
Practicalities
US OTC pricing as of 2024: magnesium glycinate ~$15–30 for 3 months; L-theanine ~$15–25 for 3 months; melatonin ~$10–20 for 6+ months; glycine powder ~$15–25 for 2–3 months at 3 g/day; apigenin ~$25–40 for 1–2 months (the most expensive). Annual cost of a full stack: ~$100–200. Quality-control variation makes brand selection material — USP-verified, NSF-certified, or third-party-tested (ConsumerLab, LabDoor) brands are more reliably accurate.
Regulatory landscape: all five are unrestricted OTC supplements in the US. Melatonin is prescription-only in the EU, UK, Japan, Australia, and Canada (Canada permits OTC). Magnesium, L-theanine, glycine, and apigenin are food / dietary-supplement category worldwide.
History
Melatonin was identified in 1958 (Lerner et al.) from bovine pineal extract; the chronobiotic role was clarified by the 1980s (Lewy et al.). OTC sale in the US dates to 1994 with the Dietary Supplement Health and Education Act (DSHEA), which categorised melatonin as a dietary supplement rather than a drug — a status not granted by most other regulators. Magnesium's role in sleep was inferred from the magnesium-deficient diet literature in the 1970s. L-theanine was isolated from green tea by Sakato in 1949 in Japan. Glycine's sleep effect was characterised by Bannai and colleagues at Ajinomoto in the 2000s. Apigenin's GABA-A binding was characterised by Viola and Wolfman in 1995; chamomile use as a sleep tisane predates the chemistry by millennia.
Audience
Older adults (60+) are the population with the strongest case for several of these. Endogenous melatonin secretion declines with age, making the physiologic-replacement logic of low-dose or prolonged-release melatonin most defensible in this group — the Lemoine 2007 prolonged-release trial and Zhdanova 2001 low-dose trial both used age-restricted populations Lemoine 2007, Zhdanova 2001. Dietary magnesium intake also declines with age and reduced caloric intake, and the Abbasi 2012 effect was demonstrated in elderly subjects Abbasi 2012.
Shift workers and jet-lagged travellers are the population with the cleanest melatonin case: chronobiotic use, not sedative use, at 0.3–0.5 mg timed to the desired phase.
Pregnant and breastfeeding women have minimal safety data across all five and should default to behavioural / hygiene approaches.
Children: pediatric melatonin ingestion has risen sharply Lelak 2022; pediatric use is appropriate only on pediatric sleep-clinic guidance (most often for autism / ADHD-associated sleep onset issues), not as casual parental dosing.
Stakes and payoff
The stakes of unsupplemented poor sleep are well-characterised — cardiovascular, metabolic (Mason 2022 documents the glucose-tolerance hit of even acute circadian disruption), cognitive, mood — but those stakes belong to the sleep-debt and circadian entries, not to a supplement entry Mason 2022. The stakes specific to this entry are narrower: choosing among supplements wrongly (over-dosed melatonin, oxide magnesium, sub-therapeutic apigenin) wastes money and creates false-negative impressions that mask real fixes. The payoff is correspondingly narrow: when the substance and dose match the actual sleep problem, sleep-onset latency falls by 7–17 minutes and subjective sleep quality improves modestly within a week or two. This is real, but it is not transformation.
The credibility range
The optimist case
Several of these supplements have plausible mechanism plus consistent (if modestly-sized) RCT evidence in their target populations. Low-dose melatonin is one of the best-studied "supplements" by trial count, with meta-analytic effects on sleep latency, total sleep time, and circadian phase that replicate across decades and across populations Ferracioli-Oda 2013, Brzezinski 2005. Magnesium has Abbasi 2012's positive controlled trial plus a coherent mechanism in deficient elderly. L-theanine has Hidese 2019's clean placebo-controlled trial showing PSQI improvement. Glycine has three converging trials and a thermoregulatory mechanism independently established in temperature-cycling sleep research. Safety profiles are excellent across the four (with apigenin the partial exception). The OTC accessibility and low cost are real advantages — total stack < $200/year vs hundreds to thousands for prescription alternatives. Used as the lower rung of a stepped-care ladder, with proper dose discipline (low-dose melatonin specifically) and recognition that supplements are adjuncts not substitutes for CBT-I and sleep hygiene, these substances earn a defensible place in the sleep toolkit.
The skeptic case
Effect sizes are uniformly small. The largest melatonin meta-analysis shows ~7 minutes of faster sleep onset — meaningful at the population level, near placebo magnitude at the individual level Ferracioli-Oda 2013. Magnesium's evidence base is three small trials with high heterogeneity and "low" GRADE certainty Mah & Pitre 2021. Glycine's literature is essentially three Japanese trials from one research group with industrial sponsorship (Ajinomoto, glycine manufacturer). Apigenin lacks any human RCT for sleep outcomes. The AASM clinical practice guideline weakly recommended against melatonin for primary insomnia AASM 2017. Quality-control failures are documented and material: Erland 2017 and Cohen 2023 show OTC melatonin contents off by –83% to +478% of label Erland & Saxena 2017, Cohen 2023. Pediatric ingestions are rising sharply with severe outcomes Lelak 2022. The first-line evidence-based treatment for chronic insomnia is CBT-I with effect sizes that dwarf any supplement Edinger 2021, Qaseem 2016; using supplements as a substitute for CBT-I is the catalogue-relevant harm. The category is also the cleanest example in consumer health of expectation effects compounding into perceived efficacy — PSQI-based outcomes are notoriously vulnerable to demand characteristics.
The author's call
This entry lands roughly between the two cases. Low-dose melatonin (0.3–0.5 mg) for circadian use, magnesium glycinate for users with low dietary intake, glycine 3 g for users who tolerate the protocol, and L-theanine for stress-driven sleep onset all have enough mechanism + evidence to be defensible occasional or short-course tools. Apigenin is markedly weaker — the human evidence does not support the specific apigenin-50-mg-for-sleep protocol that has been popularised. None of these substances belongs ahead of CBT-I and sleep hygiene in a chronic insomnia plan; they are adjuncts and short-course aids, not foundations. The score that follows reflects this: real but small sleep effect, low cost, low effort, modest evidence overall, moderate controversy concentrated in melatonin (chronic use, OTC quality control, pediatric ingestion).
Stakeholder and incentive map
- Supplement manufacturers — large commercial incentive. The US sleep-aid supplement market is estimated at $1B+ annually; the gummy-melatonin subcategory grew >50% during the 2020–2022 sleep-anxiety period. Branding pushes higher-dose products (more visible "strength") and consolidates novelty supplements like apigenin into the category.
- Sleep-medicine clinicians and AASM — professional skepticism, anchored on the 2017 guideline that recommended against melatonin for primary insomnia AASM 2017. The behavioural-treatment guideline strongly recommends CBT-I as first line Edinger 2021. Specialist field is broadly supplement-skeptical, more so for chronic use.
- Podcast / longevity community — strong commercial and cultural incentive to recommend supplements. Apigenin specifically rose in the Huberman / Attia podcast ecosystem in 2021–2023 with limited human evidence underneath. Magnesium glycinate became a near-universal recommendation in this space.
- FDA — regulates OTC supplements post-market under DSHEA; cannot pre-approve content. The pediatric melatonin surge and the JAMA gummy-dose-deviation study have prompted increased post-market scrutiny but no structural reform.
- EU / UK / Japan / Australia / Canada (partial) regulators — treat melatonin as a prescription medicine. Substantive disagreement with US framing.
- Consumer-testing labs — ConsumerLab, LabDoor, USP, NSF — counter-incentive: publishing accuracy data on supplement contents. Their reports drive the practical brand-selection advice for the protocol section.
Population variability
Strong responders are characterised by some combination of: low dietary magnesium intake (magnesium responders), measurably reduced endogenous melatonin secretion (older adults — melatonin responders), high pre-sleep autonomic arousal / anxiety (L-theanine and glycine responders, possibly apigenin), genuine circadian misalignment (low-dose melatonin responders), and an otherwise well-disciplined sleep environment so the supplement is being asked to do something small rather than something foundational.
Weak responders: chronic insomnia of moderate-severe presentation (effect sizes too small for clinical change), untreated obstructive sleep apnea (no supplement fixes airway collapse), alcohol-impaired sleep (alcohol disrupts sleep architecture more than any supplement can compensate), severe shift-work disorder (often requires structural schedule fix), and clinical anxiety / depression where the sleep complaint is a downstream symptom.
Gender / age: melatonin secretion declines markedly with age (60+ skews most responsive). Perimenopausal women have sleep complaints that often track with vasomotor symptoms — supplements address the wrong target. Pediatric pharmacokinetics differ enough that adult dosing extrapolation is unsafe. Pregnancy data is sparse to absent across the category.
Knowledge gaps
- Long-term safety (>1 year). Almost all trials are 4–12 weeks. The melatonin literature's longer-term studies (Andersen 2016 reviews available safety data) suggest no major signal but the surveillance is patchy Andersen 2016.
- Combination effects. The "sleep stack" (magnesium + glycine + theanine + low-dose melatonin) is widely used but never RCT-tested as a combination.
- Apigenin in humans for sleep specifically. No controlled human trials of purified apigenin for sleep outcomes. The current dose (50 mg) is extrapolated, not validated.
- Tolerance / dependence. Subjective tolerance is reported in long-term melatonin users (effect "fading after weeks"); no rigorous trial characterises this. Other four substances have no clear tolerance signal but the duration of follow-up is short.
- Magnesium form head-to-head. Glycinate vs threonate vs citrate has not been compared head-to-head in a sleep RCT despite confident clinical advice favouring glycinate.
- Quality control downstream. Better OTC content accuracy data exists for melatonin than for the others — the apigenin and glycine markets are essentially unmeasured.
- Endogenous-melatonin restoration vs suppression. Whether chronic exogenous melatonin suppresses endogenous secretion or downregulates receptors in long-term users is mechanistically plausible but not rigorously characterised in humans.
Coverage relative to brief. The brief named magnesium, L-theanine, low-dose melatonin, glycine, and apigenin, with consequences spanning sleep onset, continuity, next-day grogginess, and chronic-vs-occasional use. All five substances are covered end to end; the four consequences are addressed in the evidence, protocol, misconceptions, and failure-modes sections respectively. No narrowing.
Hard rating calls.
evidence: 2reflects the median substance, not the best. Low-dose melatonin alone would justify a 3 (Ferracioli-Oda 2013, Brzezinski 2005, Auld 2017 — multiple meta-analyses) but apigenin pulls hard the other way (no human sleep RCTs), and the AASM 2017 weak-against recommendation for melatonin in primary insomnia complicates the headline. The honest pooled call is 2.controversy: 3is concentrated in melatonin (chronic use, OTC dose accuracy per Erland 2017 and Cohen 2023, pediatric ingestion per Lelak 2022, EU/UK/Japan/Australia prescription-only status) and apigenin (podcast endorsement vs absent human sleep evidence). Magnesium, L-theanine, and glycine are individually much less contested.action: dooverdecide— none of these require clinician input outside the flagged contraindications. Borderline, though; the entry is more a "decide which one" survey than an unconditional "do."cadence: as-neededhonestly reflects the entry's recommended use pattern (occasional, short-course, adjunct). Some users take chronically; that's discussed as a concern, not endorsed.- Did not score
beauty_cumulativeorlongevitydespite sleep being upstream of both — the supplement-specific effect is too small and indirect to defend a non-zero score under the evidence gate. Sleep-debt and circadian-alignment entries are the right home for those consequences.
Editorial calls during the write.
- Chose to lead with mechanism rather than CBT-I framing. A reader picking up the entry expects to learn about supplements; pushing the bigger-lever (CBT-I, hygiene) conversation to the
alternativessection lets the reader meet the substance on its own terms first, then encounters the honest hierarchy. The alternative — leading with "don't bother, do CBT-I" — risks reader bounce before the actual evidence lands. - Did not use a
stakessection. The stakes of skipping these supplements are small (the entry says so honestly); the stakes of misusing them are covered inmisconceptionsandcontraindications. A dedicated stakes section would have inflated the consequence of inaction. - Apigenin gets honest, sustained skepticism throughout — chose not to drop it from the entry (it's part of the brief and the consumer reality) but called the evidence gap plainly. Risk: reader expecting validation of a popularised supplement may bounce. Editorial judgement: trust beats hype.
Future-link candidates. sleep-hygiene-fundamentals, morning-light-exposure (or whatever id is chosen for the light-exposure-sleep entry), cbt-i, sleep-apnea, alcohol-and-sleep, caffeine-timing. The closing out-of-scope section names these in reader-friendly form; wire them in as related meta once the ids exist.
Separate-entry candidates. CBT-I deserves its own entry in the mental or sleep category — it is the strongest sleep intervention in the catalogue and currently appears only as the "alternatives" pointer here. Pediatric melatonin safety also warrants either its own short entry or a callout in a broader pediatric-supplements piece — the Lelak 2022 surge is large enough that "talk to a pediatric sleep clinician" deserves more than the warning bullet it gets here.
Known dossier-vs-article delta. The dossier carries citations not used in the article body (Andersen 2016 on long-term melatonin safety, Mason 2022 on circadian/glucose, Boyle 2016 on magnesium anxiolysis, Salehi 2019 on apigenin pharmacology beyond the receptor binding the body uses, Kawai 2015 mechanism detail). All article cites resolve in the dossier.
Sleep Supplements
Typical OTC pricing: ~$15–40 per supplement per 1–3 months in the US. A single supplement runs under $50/year; a full stack stays under $200/year. Among the cheapest interventions in the catalogue.
One capsule or scoop, 30–60 minutes before bed. No protocol complexity, no preparation; daily compliance is the only ask.
Clear, replicated improvement in sleep-onset latency: ~7 minutes for melatonin in meta-analysis (Ferracioli-Oda 2013), ~17 minutes for magnesium in elderly (Mah & Pitre 2021), PSQI improvement for L-theanine (Hidese 2019), shortened PSG sleep onset for glycine (Yamadera 2007). Effects are small individually but consistent across substances and trials.
Modest day-to-day wellness lift via improved sleep onset and continuity. Sleep is upstream of how the next day feels; magnesium and low-dose melatonin meta-analyses show small but real sleep gains that propagate forward (Mah & Pitre 2021; Ferracioli-Oda 2013). Effect is indirect and small in repleted, otherwise-healthy adults.
Knock-on daily-energy improvement from better sleep. Yamadera 2007 and Bannai 2012 documented reduced next-day sleepiness and improved PVT after glycine 3 g pre-bed, including after partial sleep restriction. Indirect, modest, but consistent.
L-theanine produces acute anxiolytic alpha-wave EEG signature and reduces subjective stress (Hidese 2019); magnesium has documented anxiolytic effects (Boyle 2016). Mood lift is partly direct (calming pre-sleep) and partly indirect (better sleep → steadier next-day mood).
Sparse-to-contested overall. Melatonin has the strongest base (multiple meta-analyses, Ferracioli-Oda 2013, Brzezinski 2005) but small effect sizes and AASM 2017 weakly recommends against it for primary insomnia. Magnesium meta-analysis (Mah & Pitre 2021) reports low GRADE certainty. Glycine rests on three small Japanese trials from one group. L-theanine has clean but small RCTs. Apigenin has no human sleep RCTs.
Small next-day cognitive improvement after better sleep onset; L-theanine has weak acute attention effects (Hidese 2019). Not the reason to take these — focus benefit is downstream of the sleep benefit.