Sleep & Anxiety Supplements: What Works, What Doesn't

This page covers the supplements people reach for to sleep and to calm anxiety: magnesium (glycinate and L-threonate), L-theanine, melatonin, apigenin, ashwagandha, glycine, oral GABA, phosphatidylserine, taurine, valerian, and CBD (cannabidiol). The honest summary: most are modest at best, none replace cognitive behavioural therapy for insomnia (CBT-I) for chronic insomnia, but a few have a reasonable evidence base and a favourable safety profile. This is also the page that explicitly addresses safety, tolerance, and addiction — the most common questions about natural sleep aids.

Magnesium glycinate (200–400 mg)

Evidence: Weak. Real, small, and low-certainty — see the trial detail below.

Magnesium calms the nervous system through two mechanisms: it blocks the excitatory NMDA (N-methyl-D-aspartate) glutamate receptor — one of the brain's main "go" signals — and acts as a cofactor for GABA (gamma-aminobutyric acid), its main "stop" signal.^[1] About half of adults in wealthy countries fall short of the recommended dietary intake, and population data show a U-shaped relationship between magnesium intake and sleep duration — both deficiency and excess track with worse sleep.

The form matters more than for almost any other supplement. Bisglycinate — magnesium bound to two glycine molecules — is absorbed through dipeptide channels rather than the usual mineral route, so it causes less of the osmotic diarrhoea seen with citrate or oxide, and the glycine itself is mildly calming.

• A 2025 randomised trial in 155 adults reporting poor sleep found that 250 mg/day of elemental magnesium from bisglycinate reduced the Insomnia Severity Index — a 0–28 questionnaire score on which anything above 15 counts as moderate insomnia — by about 1.6 points more than placebo over four weeks.^[2] Two things temper that. The effect was small on the trial's own reckoning (a standardised effect of about 0.2) and the result cleared the significance threshold by a hair. And most people did not benefit: the trial counted a drop of six or more points on the Insomnia Severity Index as a clinically meaningful improvement, and only about 30% of participants reached it — 26 people on magnesium against 15 on placebo. That is eleven extra responders among the 155 people enrolled — very roughly one additional person in seven treated who got a meaningful benefit they would not have had on placebo. The clearest responders were adults with low dietary magnesium to begin with.
• The wider trial base behind magnesium-for-sleep is tiny: the meta-analysis usually cited for it pooled three randomised trials, and rated the evidence low-certainty by GRADE, the standard system for grading evidence quality.^[3] A larger synthesis across nine studies and 7,582 people makes the split that matters: observational data link higher magnesium intake to better sleep, while the randomised trials contradict each other.^[4] That is why this page rates the sleep effect Weak rather than Moderate: the signal is genuine, but it rests on one barely-significant trial with a small effect, not on a consistent body of them.
• A smaller crossover trial reported improvements in subjective sleep quality and heart rate variability, but it remains an unpublished preprint that has not been peer-reviewed, and its authors have supplement-industry ties — so it carries little weight next to the trial above.^[5]

Safety:

• Very well-tolerated; bisglycinate is the gentlest form on the gut.
• Kidney disease is a contraindication for high-dose supplementation.
• Separate by at least two hours from quinolone and tetracycline antibiotics and from bisphosphonates — magnesium binds them.
• The European Food Safety Authority (EFSA) upper limit for supplemental magnesium is 250 mg/day; the US upper limit is 350 mg/day. The commonly recommended 200–400 mg range sits at or above the EFSA limit, so the upper end is not "more is better" territory.

Addiction / tolerance / dependence: None documented. No tolerance, no withdrawal, no rebound insomnia. This is a safe long-term option.

Practical: 200–400 mg elemental magnesium as bisglycinate, 1–2 hours before bed. Start at the lower end: 250 mg is both the dose used in the trial above and the EU supplemental limit, and there is no evidence the top of the range works better.

Magnesium L-threonate (the cognition-and-sleep-architecture form)

Evidence: Moderate for cognition; Weak-to-moderate for sleep.

L-threonate is a magnesium form engineered to cross the blood-brain barrier. The claim traces to a single foundational rodent study: chronic oral magnesium L-threonate raised magnesium concentrations in the cerebrospinal fluid of rats and enhanced synaptic plasticity, learning, and memory — an effect that other magnesium salts did not produce.^[6] That is a real result, but note what it is: rats, not people. There is no published human study directly measuring cerebrospinal-fluid magnesium after L-threonate, so "it crosses the blood-brain barrier in humans" is an extrapolation, not a measured fact.

• A six-week trial in 100 adults aged 18–45 who were dissatisfied with their sleep found that 2 g/day of the branded L-threonate ingredient improved working memory on a standard cognitive battery, with the authors translating the overall gain into an estimated 7.5-year difference in cognitive performance.^[7] The trial was funded by Threotech, which owns the ingredient's intellectual property and helped design the study, and both authors work for a contract research organisation funded by nutraceutical companies — the same author, in fact, is on the bisglycinate trial cited above, so the two magnesium results are not independent. Only one of seven subtests reached significance and neither constituent composite did, yet the combined score that carries the "brain age" figure did. Treat that figure with caution — it is a surrogate readout from a single trial, not a replicated hard outcome, and the same trial found no benefit on a reasoning test or on any of the objective sleep measures from participants' sleep-tracking rings.
• On sleep, a 21-day randomised placebo-controlled trial in 80 adults aged 35–55 with self-reported sleep problems found total sleep time roughly unchanged but better deep-sleep and REM (rapid-eye-movement) sleep scores on an Oura ring, plus improved mood and daytime energy.^[8] The heavy caveat: several authors are employed by the ingredient manufacturer, outcomes were consumer-wearable and questionnaire-based rather than polysomnography (overnight sleep-lab recording, the reference standard), and the paper has since had a published correction issued against it. Read it as manufacturer-run and preliminary. The "better deep-sleep scores" endpoint in particular is close to uninterpretable — consumer rings and watches miss roughly half of true deep-sleep epochs, and across the devices tested their deep-sleep totals ran anywhere from about 25 minutes short to more than 44 minutes long.^[9] A change in a wearable's deep-sleep output is not evidence of a change in deep sleep. See Sleep architecture.

Practical: L-threonate is the form to consider if cognition or night-time mental "racing" is the issue rather than physical tension; bisglycinate remains the cheaper, better-evidenced choice for general sleep. The two are sometimes stacked (threonate earlier, bisglycinate before bed) — but do the arithmetic first, because the threonate contribution is not small. Magnesium L-threonate is only about 7% magnesium by weight, so the 2 g dose used in trials delivers 145 mg of elemental magnesium, well over half the European supplemental limit on its own.^[10] Stacked on 250 mg from bisglycinate that is nearly 400 mg a day, and at the top of the 200–400 mg range above it is over 500 mg — past both limits quoted earlier. If you stack, lower one of the two.

L-theanine (200–450 mg)

Evidence: Weak to Moderate.

L-theanine is an amino acid found in tea that crosses into the brain and promotes a state of relaxed alertness without sedation — it raises alpha brain-wave activity and modulates glutamate and GABA rather than switching them off.^[11]

• A narrative review of 13 trials in 550 participants concluded that 200–450 mg/day improves sleep onset, efficiency, and feeling refreshed without next-day grogginess — though it pooled nothing, and three of its four authors work for a tobacco company.^[12] The actual meta-analysis, across 19 articles and 897 people, finds small effects and finds them all on subjective measures: about 0.15 of a standard deviation on falling asleep, 0.33 on daytime functioning, 0.43 on sleep quality.^[13]
• The anxiety case is weaker than it is usually presented. The single generalised-anxiety-disorder trial inside the review most often cited for it found no benefit on anxiety symptoms,^[14][15] and a 2026 meta-analysis of 31 trials reached the same conclusion for clinical anxiety — while confirming a real short-term gain in attention and a modest reduction in acute stress.^[16] L-theanine's better-supported use is the everyday one: calmer alertness, not treatment for an anxiety disorder.
• An earlier 2019 crossover trial in 30 healthy adults (200 mg/day for four weeks) reported lower stress-related symptoms and better sleep.^[17]

Safety: Well-tolerated across trials; the US Food and Drug Administration (FDA) grants it GRAS (generally recognised as safe) status. A theoretical additive blood-pressure-lowering effect means caution if you take several antihypertensives. Long-term (>1 year) data is limited but shows no harm signal.

Addiction / tolerance / dependence: None documented. The mechanism is non-sedative, so there is no rebound insomnia or anxiety on stopping.

Practical: A standard cup of tea has only ~25 mg and comes with caffeine, so a supplement is needed to reach the therapeutic range. 200 mg in the evening; up to 200–400 mg/day if you want the calm-alertness effect during the day — not as a treatment for an anxiety disorder, where the evidence is null. Usually safe to combine with magnesium glycinate.

Melatonin (0.3–1 mg) — and a new long-term safety question

Evidence: Moderate for narrow indications; weak for general insomnia. New caution on chronic high-dose use.

Melatonin is the hormone the pineal gland releases in darkness. The decline with age is smaller than the folklore suggests: when healthy, drug-free older adults are measured under controlled conditions, their overnight melatonin is indistinguishable from young adults'.^[18] As a supplement it is best supported for circadian problems — jet lag, delayed sleep-phase syndrome, shift-work transitions^[19] — and for insomnia in older adults, where the guideline evidence is strongest, though the rationale is empirical rather than replacement of a deficiency. For ordinary insomnia in younger and middle-aged adults the effect is small, and two meta-analyses agree on how small. The larger, covering 19 trials and 1,683 people, found melatonin cut the time to fall asleep by about seven minutes and added about eight minutes of total sleep;^[20] an earlier one across 17 trials found roughly four minutes faster onset and thirteen minutes more sleep.^[21] Either way the gain is a matter of minutes, not hours. The 2017 American Academy of Sleep Medicine guideline on sleep drugs gave a conditional recommendation against melatonin for sleep-onset or maintenance insomnia in adults.^[22]

Dose: For shifting the clock, low doses (0.3–1 mg) timed correctly are what the evidence supports. For falling asleep the picture is different and the article's older "low is always better" framing is too simple: a dose-response pooling of 26 randomised trials puts the sleep-onset optimum near 4 mg, given about three hours before bed rather than the usual half-hour.^[23] Higher still — 10 mg — is not supported by either target, and larger doses are more likely to leave you groggy the next day.

The new cardiovascular signal. An observational analysis compared about 65,000 long-term melatonin users (a year or more) against matched non-users and reported higher rates of heart failure, heart-failure hospitalisation, and death over five years. In absolute terms it reported heart failure in about 4.6% of melatonin users against 2.7% of non-users, and death in about 7.8% against 4.3% — roughly nineteen extra heart-failure cases and thirty-five extra deaths per 1,000 users over five years, if the association were causal. Weight this carefully: it is a conference abstract presented at the American Heart Association's Scientific Sessions 2025, not a peer-reviewed paper — AHA states plainly that its meeting abstracts are not peer-reviewed and the findings are preliminary until published in full.^[24] The underlying methods have not been subjected to external review, so the effect size in particular should not be quoted as established — and as of mid-2026, eight months after presentation, it still has not appeared as a full paper. It is also observational data and cannot establish cause — the people who take melatonin nightly for years tend to have worse underlying insomnia, more depression, and undocumented use of other sleep drugs, any of which could drive the association.^[25] One further limitation matters for how you read it: the database mixes countries where melatonin is prescription-only with countries where it is sold over the counter, and the researchers could not tell which was which. Anyone buying it off the shelf would have been counted as a non-user, which the authors flag as a reason the analysis may not reflect reality. Some of these users did have insomnia severe enough to be medicated for a year or more — but the comparison is not clean in either direction. But the signal is large enough to argue against treating melatonin as a benign substance to take indefinitely at high doses. The conservative reading: use it strategically — for circadian resets, at the lowest effective dose — rather than as an open-ended nightly habit.

Melatonin's appeal beyond sleep rests on its role as a mitochondrial antioxidant, and some have proposed anti-ageing benefits on that basis;^[26] a 2026 meta-analysis in postmenopausal women found a modest gain in bone mineral density but no improvement in anxiety or in hot flushes and night sweats.^[27] One genotype-specific signal is worth noting rather than generalising: in a randomised trial, 5 mg of melatonin worsened glucose tolerance in carriers of a common MTNR1B risk variant and not in anyone else.^[28] Across the wider trial literature — 31 randomised trials — melatonin slightly improved markers of insulin resistance, so this is a caveat for one subgroup, not a metabolic harm claim.^[29] None of this overrides the everyday point: low dose, narrow indication, not forever.

Quality: in a chromatographic analysis of 31 commercial supplements, actual melatonin content ranged from 83% below to 478% above the labelled dose, more than 71% missed their label claim by over 10%, lot-to-lot variation within a single product reached 465%, and serotonin was detected in eight of the products.^[30] A 2023 analysis of 25 melatonin gummies found the same problem in the format most people now buy: melatonin ranged from 74% to 347% of the labelled quantity, 88% were inaccurately labelled, and one product contained no detectable melatonin at all but 31.3 mg of CBD. Notably, serotonin was not detected in any of those gummies — that finding belongs to the older Canadian capsule sample and did not replicate.^[31] The EU prolonged-release prescription product (Circadin 2 mg) is far more reliably dosed — and it is more than a purity story: it holds a licensed indication for insomnia in adults aged 55 and over, which the European insomnia guideline endorses for up to three months.^[32]

Addiction / tolerance / dependence: None; no withdrawal.

Apigenin (around 50 mg)

Evidence: Weak / preliminary.

Apigenin is the flavonoid in chamomile (and parsley and celery) that has become a popular bedtime supplement, usually at a 50 mg nightly dose. In rats it acts as a gentle positive modulator of the GABA(A) receptor — enhancing the brain's own inhibitory signalling without the amnesia or motor impairment of benzodiazepines; that mechanism has not been demonstrated in people.^[33] Its safety record at typical doses is good, but direct human trials of isolated apigenin for sleep at 50 mg are thin — most of the support is extrapolated from whole-chamomile studies and animal work.^[34]

The longevity interest in apigenin comes from a separate mechanism: it inhibits CD38, an enzyme that consumes NAD⁺ (nicotinamide adenine dinucleotide, a coenzyme central to energy metabolism and DNA repair) and whose activity climbs with age. In animal models this preserves NAD⁺ and dampens inflammatory cytokines.^[35] That is a genuinely interesting bridge between sleep and ageing biology, but it remains preclinical — it is not a reason to expect anti-ageing effects in people at the doses sold for sleep.

Practical: Reasonable to try at 50 mg for mild sleep onset difficulty; set expectations to "gentle," and don't buy the NAD⁺ longevity framing as established.

Ashwagandha (300–600 mg standardised root extract)

Evidence: Moderate for short-term stress and anxiety; Moderate for sleep.

Withania somnifera is an adaptogen that lowers cortisol by buffering the body's stress axis (the hypothalamic-pituitary-adrenal, or HPA, axis). A 2026 meta-analysis reported a dose-dependent reduction in stress, anxiety, and depressive symptoms in adults — the population was not restricted to healthy volunteers —^[36] and several small trials show improved sleep quality and shorter sleep onset, particularly in stressed or insomniac groups.^[37] The US National Institutes of Health (NIH) Office of Dietary Supplements rates the sleep evidence as suggestive.^[38] Animal work also suggests it can restore the age-blunted rhythm of circadian "clock genes": in aged rats, Withania somnifera leaf extract partially restored the daily rhythms of two of three regulators that normally cycle with the clock — a longevity-associated repair protein and the master switch for the cell's own antioxidant defences — in the suprachiasmatic nucleus, the brain's master clock; the core clock gene tested was insensitive to it.^[39] If that held in humans it would be more interesting than simple sedation — but it is a rat study, and it used the leaf extract, not the root extract recommended below.

Extract matters. Most quality trials use one of two standardised extracts. KSM-66 is a root-only extract (standardised to ~5% withanolides — the steroidal compounds that carry the plant's activity), dosed 300–600 mg/day, with the largest human safety and efficacy record.^[40] Sensoril is a more concentrated root-and-leaf blend dosed lower (125–250 mg/day); the leaf component is worth noting because leaf extracts have shown liver toxicity in laboratory cell studies, whereas standardised root extract has a clean safety profile.^[41] Prefer a standardised root extract.

Cautions:

• Liver injury — the case reports are real, and the reassuring half of this caveat does not hold up. A 2026 scoping review of 25 published cases found no consistent dose-response and no basis for pinning the cases on non-standardised or leaf-containing products, largely because formulation and plant part usually go unrecorded.^[42] In the largest series the implicated products were chemically clean, single-ingredient extracts.^[43] Injury is uncommon and usually resolves on stopping, but a reputable brand is not a guarantee. Denmark banned ashwagandha in food supplements in 2023.^[44]
• Avoid in thyroid disease (it can stimulate the thyroid), autoimmune conditions, and pregnancy.
• It is sedating: do not combine with benzodiazepines, Z-drugs (zolpidem and relatives), opioids, or alcohol, where effects are additive.^[45]
• It has mild blood-pressure- and blood-sugar-lowering effects (monitor if on the relevant medications) and modulates the liver enzyme CYP3A4, which metabolises a large share of prescription drugs — a route to interactions.^[46]

Addiction / tolerance / dependence: None documented, but long-term safety data is limited.

Glycine (3 g before bed)

Evidence: Weak.

Glycine is both an amino acid and an inhibitory neurotransmitter. A 3 g dose before bed has been shown in small trials to improve subjective sleep quality, shorten sleep onset, and improve next-day alertness.^[47] The mechanism, worked out in rats, is thermoregulatory rather than sedative: glycine acts on NMDA receptors in the suprachiasmatic nucleus, raising cutaneous blood flow so the body sheds heat and core temperature falls — the physiological trigger for sleep onset.^[48] The honest caveat: the human trials are small, and essentially all of this work — mechanism and clinical — comes from one research group at a single amino-acid manufacturer. There is no independent replication and no meta-analysis.

One thing to notice if you are also taking magnesium bisglycinate: the flagship magnesium trial delivered 250 mg of magnesium and about 1.5 g of glycine daily, so the two are not cleanly separable, and someone stacking both is taking roughly 4.5 g of glycine a day.

Safety: Very well-tolerated. Not addictive. A low-cost, low-risk option to try.

GABA supplements (why oral doses mostly don't work)

Evidence: Weak / largely negative.

GABA is the brain's primary inhibitory neurotransmitter, which makes oral GABA supplements intuitively appealing — but ingested GABA barely crosses the blood-brain barrier, so most of it acts on peripheral tissues or is broken down before reaching the brain.^[49] Some studies of fermented or specially formulated GABA report small subjective reductions in sleep latency, but objective sleep measures generally show no significant benefit over placebo.^[50] Skip it in favour of compounds that actually reach the brain (magnesium, L-theanine, glycine, apigenin).

Phosphatidylserine (100–400 mg as sold)

Evidence: Weak / preliminary.

Phosphatidylserine is a phospholipid concentrated in nerve-cell membranes; sunflower- or soy-derived supplements have been studied for blunting the cortisol response to stress and for cognition.^[51] A 2025 randomised controlled trial of 100 mg/day of sunflower-derived phosphatidylserine in healthy children was null on its primary and secondary outcomes; only a pre-defined subgroup of below-median performers improved on a visuospatial memory task.^[52] For sleep and anxiety in adults there is essentially nothing: two full PubMed searches this year turned up no controlled trial of phosphatidylserine for either. It may also reduce the effect of anticholinergic medications. On current evidence there is no basis for recommending it for sleep at all.

Taurine (the longevity-molecule cautionary tale)

Evidence: Weak, and not for sleep. The anti-ageing case has weakened.

Taurine became a longevity story in 2023 when a high-profile study extended the healthy lifespan of middle-aged mice by up to 12% and reported that taurine levels fall with age.^[53] Human data since then has undercut the premise: a longitudinal analysis across human and primate cohorts found that circulating taurine does not reliably decline with age — it often stays flat or rises, and varies far more between individuals than across age groups — making it a poor biomarker of human ageing.^[54] Taurine has legitimate uses in cardiovascular and exercise contexts, but it is neither a sleep aid nor a substantiated human longevity intervention. Included here only because it is heavily marketed alongside the others.

Valerian

Evidence: Weak / mixed.

Valerian root acts on GABA signalling via valerenic acid. An umbrella review of eight systematic reviews found a good safety profile but no demonstrated efficacy for insomnia: valerian does appear to improve subjective sleep quality, and that improvement has never been reproduced on quantitative or objective measures.^[55] The earlier meta-analytic literature was already mixed in the same way.^[56]

Valerian avoids the dependence of benzodiazepines and Z-drugs. The widely repeated advice to cycle it (a few weeks on, a 1–2 week break) to avoid a blunted effect is a folk heuristic rather than a finding: no trial has tested tolerance to valerian, and there is no controlled long-term safety data at all — the review by the US National Center for Complementary and Integrative Health (NCCIH) notes that studies have been short and long-term safety is unknown.^[57] Side effects are mild — occasional morning grogginess, vivid dreams. It is contraindicated in pregnancy, in children, and alongside other sedatives or central-nervous-system depressants. Given the mixed objective evidence, it is reasonable to skip in favour of better-supported options.

CBD (cannabidiol)

Evidence: Weak. Preliminary and inconsistent.

Sleep effects in trials are mixed, most over-the-counter products have poor quality control, and the only CBD product approved by the US Food and Drug Administration (Epidiolex) is for specific seizure disorders, not sleep. Cautions: CBD inhibits the liver enzymes CYP3A4 and CYP2C9, creating interactions with statins, warfarin, and many other drugs;^[58] liver-enzyme elevations also show up at doses in the consumer range — in a randomised trial run by the FDA, 28 days at about 350 mg a day for a 70 kg adult pushed liver enzymes past three times the normal ceiling in roughly one healthy adult in twenty, though that is the top of what people take, not a typical gummy dose.^[59] Skip absent a specific indication.

The addiction question, directly

Evidence: Moderate for the absence of dependence in supplements; Strong for the dependence risks of the prescription drugs they are compared against.

None of the supplements above produces clinically significant addiction, tolerance, or withdrawal at the doses studied. That is their main advantage over prescription hypnotics:

• Benzodiazepines, Z-drugs, and over-the-counter antihistamines (diphenhydramine — Benadryl, ZzzQuil) all carry meaningful tolerance, dependence, and withdrawal. The benzodiazepines and Z-drugs also carry a hip-fracture signal,^[60] and heavy cumulative use of anticholinergic drugs — the class the antihistamines belong to — tracks with higher dementia risk in adults over 65.^[61] A 2025 microsimulation is the source that actually speaks to adults over 50: it estimated that if the 15.3 million Americans over 50 on prescription sleep medication stopped, lifetime falls would fall by 8.5% and cognitive impairment by 2.1%.^[62] See Treating chronic insomnia for the full picture.
• Dual orexin-receptor antagonists (suvorexant, lemborexant, daridorexant) — which block orexin, the brain's wake-promoting signal, rather than sedating you — show no tolerance, withdrawal or rebound insomnia when stopped, a real advantage over the older drugs. They are nonetheless Schedule IV controlled substances: in studies of people who use sedatives recreationally they produce measurable drug-liking — less than zolpidem at the dose actually prescribed, but more than placebo.^[63] Real-world abuse reports remain very low.^[64]

Three honest caveats. "No chemical addiction" doesn't rule out a psychological "I can't sleep without it" reliance, which can attach to any sleep aid. The absence of withdrawal has also never actually been tested: the trials behind these supplements run three to six weeks with no discontinuation phase, so it is an untested assumption rather than a finding. And long-term (>1–2 years of continuous daily use) safety data is limited for most of these — magnesium and, with the new cardiovascular caveat, melatonin have the longest track records.

Supplements versus CBT-I: not the same league

Evidence: Strong.

It is worth being blunt about magnitude. Even the best-supported supplement here (magnesium bisglycinate) produces a small effect on insomnia. Cognitive behavioural therapy for insomnia (CBT-I) produces large effects and the gains persist after treatment ends — and it is the treatment the guidelines put first, ahead of any drug.^[65] No head-to-head trial has put a number on "several times better," so read the comparison as one of durability rather than of measured magnitude: drug effects stop when the drug stops. CBT-I works because it targets the actual machinery of chronic insomnia — the conditioned arousal and sleep anxiety that supplements don't touch.

Despite that, it is wildly underused. Sweden is the clearest illustration: in 2024 roughly 870,000 people collected a sleeping-pill prescription, against about 600 recorded as receiving internet-delivered CBT-I. Those figures come from a Karolinska Institutet news item citing national registry counts rather than from a peer-reviewed study, and the CBT-I number is an undercount — it captures only the internet-treatment registry, not therapy delivered through health centres.^[66] Even allowing for that, the order-of-magnitude gap is the point. The right framing is to use supplements as occasional adjuncts that lower the physiological barrier to rest while the durable work — sleep regularity, light, and CBT-I where needed — does the heavy lifting. See Treating chronic insomnia.

Why this matters for ageing

Evidence: Moderate for the link between disrupted sleep and faster epigenetic ageing; no evidence that any sleep supplement slows it.

The reason to fix sleep rather than just sedate it: chronic insomnia accelerates the DNA-methylation "epigenetic clocks" used to estimate biological age, with measurable shifts in markers like GrimAge and a shortening of telomere length as estimated from methylation patterns (telomeres are the protective caps on chromosomes that shorten as cells divide).^[67] Both of those findings come from cross-sectional samples of a few dozen to a few hundred people, which is thinner than the Moderate rating above implies. Genetic evidence points the same way — a 2025 Mendelian randomisation analysis found insomnia associated with faster epigenetic ageing — but this remains a suggestive literature rather than a settled one.^[68] Regular, consolidated rest-activity rhythms track with slower epigenetic ageing: in a Baltimore cohort with 7-day wrist actigraphy and blood DNA methylation, stronger and more regular 24-hour rhythms were associated with less epigenetic age acceleration on two of the four clocks tested (GrimAge and PhenoAge); the other two pointed the same way but were not statistically significant.^[69] That study is cross-sectional, so it shows association, not that fixing your rhythm winds the clock back. A supplement that genuinely improves sleep architecture could in principle slow that clock — but no sleep supplement has yet shown this in humans, which is exactly why the bar is "restore real sleep," not "mask the symptom." See Sleep for the full longevity case.

Tier ranking summary

Further reading

• Schuster J et al. Magnesium Bisglycinate Supplementation in Healthy Adults Reporting Poor Sleep: A Randomized, Placebo-Controlled Trial. Nat Sci Sleep 2025.^[70]
• Lopresti AL, Smith SJ. The effects of magnesium L-threonate (Magtein) on cognitive performance and sleep quality in adults: a randomised, double-blind, placebo-controlled trial. Front Nutr 2025.^[71]
• Cotter J et al. Examining the effect of L-theanine on sleep: a systematic review of dietary supplementation trials. Nutr Neurosci 2026.^[72]
• Sateia MJ et al. Clinical Practice Guideline for the Pharmacologic Treatment of Chronic Insomnia in Adults. J Clin Sleep Med 2017 (AASM) — the guideline that recommends against melatonin for insomnia.^[73]
• Edinger JD et al. Behavioral and psychological treatments for chronic insomnia disorder in adults: an American Academy of Sleep Medicine clinical practice guideline. J Clin Sleep Med 2021 — the guideline behind CBT-I as first-line.^[74]
• Long-term melatonin use and heart failure — conference abstract MP2306, AHA Scientific Sessions 2025; not peer-reviewed.^[75]
• Ferracioli-Oda E et al. Meta-analysis: melatonin for the treatment of primary sleep disorders. PLoS One 2013 — 19 trials, 1,683 people; the source of the "seven minutes" figure.^[76]
• Slutsky I et al. Enhancement of learning and memory by elevating brain magnesium. Neuron 2010 — the rodent basis for the L-threonate blood-brain-barrier claim.^[77]
• Hausenblas HA et al. Magnesium-L-threonate improves sleep quality and daytime functioning in adults with self-reported sleep problems: a randomized controlled trial. Sleep Med X 2024 — manufacturer-affiliated authors; corrigendum issued 2025.^[78]
• Bannai M, Kawai N. New therapeutic strategy for amino acid medicine: glycine improves the quality of sleep. J Pharmacol Sci 2012.^[79]
• Kawai N et al. The sleep-promoting and hypothermic effects of glycine are mediated by NMDA receptors in the suprachiasmatic nucleus. Neuropsychopharmacology 2015.^[80]
• Kukkemane K, Jagota A. Therapeutic effects of hydro-alcoholic leaf extract of Withania somnifera on age-induced changes in daily rhythms of Sirt1, Nrf2 and Rev-erbα in the SCN of male Wistar rats. Biogerontology 2020.^[81]
• Liu C et al. Twenty-Four-Hour Rest-Activity Rhythms and Epigenetic Age Acceleration in Middle-Aged and Older Adults. JAMA Netw Open 2026.^[82]
• Erland LA, Saxena PK. Melatonin Natural Health Products and Supplements: Presence of Serotonin and Significant Variability of Melatonin Content. J Clin Sleep Med 2017.^[83]
• NIH ODS Ashwagandha Fact Sheet for Health Professionals.^[84]
• Kramer DJ, Johnson AA. Apigenin: a natural molecule at the intersection of sleep and aging. Front Nutr 2024.^[85]
• Singh P et al. Taurine deficiency as a driver of aging. Science 2023 — the mouse-lifespan study that started the taurine story.^[86]
• NIH researchers conclude that taurine is unlikely to be a good aging biomarker. 2024.^[87]
• Valente V et al. Does valerian work for insomnia? An umbrella review of the evidence. Eur Neuropsychopharmacol 2024.^[88]
• Bulman A et al. The effects of L-theanine consumption on sleep outcomes: a systematic review and meta-analysis. Sleep Med Rev 2025.^[89]
• Gerolymos C et al. Cognitive and affective effects of L-Theanine: a systematic review and meta-analysis of 31 randomized trials. Mol Psychiatry 2026.^[90]
• Zeitzer JM et al. Do plasma melatonin concentrations decline with age? Am J Med 1999.^[91]
• Cruz-Sanabria F et al. Optimizing the Time and Dose of Melatonin as a Sleep-Promoting Drug: A Systematic Review of Randomized Controlled Trials and Dose-Response Meta-Analysis. J Pineal Res 2024.^[92]
• Cohen PA et al. Quantity of Melatonin and CBD in Melatonin Gummies Sold in the US. JAMA 2023.^[93]
• McIntyre D et al. Ashwagandha (Withania somnifera)-Associated Liver Injury: A Scoping Review of Clinical Characteristics and Safety Considerations. Cureus 2026.^[94]
• Philips CA et al. Ashwagandha-induced liver injury — a case series from India and literature review. Hepatol Commun 2023.^[95]
• Florian J et al. Cannabidiol and Liver Enzyme Level Elevations in Healthy Adults: A Randomized Clinical Trial. JAMA Intern Med 2025.^[96]
• Fernandez ME et al. Is taurine an aging biomarker? Science 2025.^[97]
• Arab A et al. The Role of Magnesium in Sleep Health: a Systematic Review of Available Literature. Biol Trace Elem Res 2023.^[98]
• He C et al. The Mechanisms of Magnesium in Sleep Disorders. Nat Sci Sleep 2025.^[99]

---