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

Most natural sleep aids deliver modest help with few side effects — but only a couple have real evidence, several popular ones are oversold, and one (long-term high-dose melatonin) now carries a cardiovascular safety signal worth knowing about. Here's the honest tier list, the addiction question, and where supplements fit alongside the one intervention that actually fixes chronic insomnia.

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, phosphatidylserine, taurine, valerian, and CBD. 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: Moderate.

Magnesium calms the nervous system through two mechanisms: it blocks the excitatory NMDA glutamate receptor and acts as a cofactor for GABA, the brain's main inhibitory 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 by about 1.6 points more than placebo over four weeks.^[2] The catch is the effect size: it was small (a standardised effect of about 0.2), and roughly 80% of participants saw no clinically meaningful change. The clearest responders were adults with low dietary magnesium to begin with.
Smaller earlier trials — including a 2024 crossover in 31 adults — showed improvements in subjective sleep quality and heart rate variability.^[3]

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 EU (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.

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 — it rides glucose transporters and measurably raises magnesium in cerebrospinal fluid, which ordinary forms do not.^[4]

A 2025 six-week trial in 100 healthy adults found that 2 g/day of magnesium L-threonate improved working and episodic memory on a standard cognitive battery, with the authors translating the gains into an estimated "7.5-year reduction in brain age."^[5] Treat that figure with caution — it is a surrogate readout from a single trial, not a replicated hard outcome.
On sleep, the same line of research found total sleep time roughly unchanged but improvements in deep- and REM-sleep scores and a rise in overnight heart rate variability, a marker of parasympathetic ("rest and digest") tone.^[6]

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 keep total elemental magnesium within the limits above — the threonate dose contributes a relatively small amount of elemental magnesium, but it still counts.

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.^[7]

A 2025 systematic review of 13 trials in 550 participants concluded that 200–450 mg/day improves subjective and objective sleep markers — onset, efficiency, and feeling refreshed — without next-day grogginess.^[8]
A separate review found L-theanine reduced symptoms in generalised anxiety disorder, schizophrenia, and ADHD when added to standard treatment.^[9]
An earlier 2019 trial in stressed adults (200 mg/day for four weeks) showed reduced anxiety and better sleep satisfaction.^[10]

Safety: Well-tolerated across trials; the 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 for daytime anxiety. 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; its output falls steeply with age. As a supplement it is best supported for circadian problems — jet lag, delayed sleep-phase syndrome, shift-work transitions — and for insomnia in older adults whose own melatonin has declined.^[11] For ordinary insomnia in younger and middle-aged adults the effect is small: meta-analysis puts it at roughly seven minutes faster sleep onset and eight minutes more total sleep.^[12] The 2017 American Academy of Sleep Medicine guideline gave a conditional recommendation against melatonin for sleep-onset or maintenance insomnia in adults.^[13]

Dose: Best evidence is for low doses (0.3–1 mg) timed correctly. Higher doses (3–10 mg) are no more effective in healthy adults and more likely to cause next-day grogginess.

The new cardiovascular signal. A 2025 cohort study presented at the American Heart Association compared about 65,000 long-term melatonin users (a year or more) against matched non-users and reported markedly higher rates of heart failure, heart-failure hospitalisation, and death over five years.^[14] This is 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.^[15] 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;^[16] a 2024–2025 meta-analysis in postmenopausal women found a modest gain in bone mineral density but no improvement in anxiety or vasomotor symptoms.^[17] None of this overrides the everyday point: low dose, narrow indication, not forever.

Quality: US over-the-counter products vary from 83% to 478% of their labelled dose.^[18] The EU prolonged-release prescription product (Circadin 2 mg) is far more reliable.

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. 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.^[19] 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.^[20]

The longevity interest in apigenin comes from a separate mechanism: it inhibits CD38, an enzyme that consumes NAD⁺ (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.^[21] 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 confirmed a dose-dependent reduction in stress, anxiety, and depressive symptoms in healthy adults,^[22] and several small trials show improved sleep quality and shorter sleep onset, particularly in stressed or insomniac groups.^[23] The NIH Office of Dietary Supplements rates the sleep evidence as suggestive.^[24] Animal work also suggests it can restore the age-blunted rhythm of circadian "clock genes," an upstream mechanism that, if it holds in humans, would be more interesting than simple sedation — but that evidence is preclinical.^[25]

Extract matters. Most quality trials use one of two standardised extracts. KSM-66 is a root-only extract (standardised to ~5% withanolides), dosed 300–600 mg/day, with the largest human safety and efficacy record.^[26] 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.^[27] Prefer a standardised root extract.

Cautions:

Rare hepatotoxicity — real case reports of liver injury exist, more often tied to non-standardised or leaf-containing products.
Avoid in thyroid disease (it can stimulate the thyroid), autoimmune conditions, and pregnancy.
It is sedating: do not combine with benzodiazepines, Z-drugs, opioids, or alcohol, where effects are additive.^[28]
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.^[29]

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 — largely by promoting peripheral blood flow that lowers core body temperature, the physiological trigger for falling asleep.^[30] The evidence base is limited and mostly from a single research group.

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.^[31] 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.^[32] Skip it in favour of compounds that actually reach the brain (magnesium, L-theanine, glycine, apigenin).

Phosphatidylserine (100–400 mg)

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.^[33] A 2026 trial in children found 100 mg of sunflower-derived phosphatidylserine improved visuospatial memory in lower-performing kids.^[34] Human sleep and anxiety data in adults is preliminary, and it may reduce the effect of anticholinergic medications. A reasonable option for stress-driven sleep disruption, but the evidence is thinner than for magnesium or L-theanine.

Taurine (the longevity-molecule cautionary tale)

Evidence: Not a sleep aid; 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.^[35] Human data since then has undercut the premise: an NIH analysis across a large longitudinal cohort 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.^[36] 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. Subjectively, 300–900 mg/day of an extract standardised to ~0.8% valerenic acid modestly improves perceived sleep quality and can shorten sleep onset by 10–20 minutes;^[37] but objective sleep-lab measures are inconsistent, and meta-analyses are mixed.^[38]

Valerian avoids the dependence of benzodiazepines and Z-drugs, but continuous nightly use can blunt its effect, so a cycling pattern (a few weeks on, a 1–2 week break) is often recommended.^[39] 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.^[40] Given the mixed objective evidence, it is reasonable to skip in favour of better-supported options.

CBD

Evidence: Preliminary and inconsistent.

Sleep effects in trials are mixed, most over-the-counter products have poor quality control, and the only FDA-approved CBD (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, and there are hepatotoxicity signals at high doses. Skip absent a specific indication.

The addiction question, directly

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 (zolpidem and relatives), and over-the-counter antihistamines (diphenhydramine — Benadryl, ZzzQuil) all carry meaningful tolerance, dependence, and withdrawal, plus a falls-and-dementia signal above age 50.^[41]
Dual orexin-receptor antagonists (suvorexant, lemborexant, daridorexant) are the newer prescription class without dependence, tolerance, or rebound — closer to the supplements in this respect.

Two 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; 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

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 — head-to-head, it is several times more effective than any supplement or hypnotic at durably restructuring sleep.^[42] 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: Swedish registry data for 2024 recorded roughly 870,000 people prescribed sleeping pills against only about 600 who received internet-delivered CBT-I.^[43] 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

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 reduction in epigenetic telomere length.^[44] Restoring regular, consolidated sleep is associated with slower epigenetic ageing.^[45] 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

Supplement	Evidence	Safety	Addiction risk	Tier
Magnesium glycinate	Moderate	Excellent	None	Worth trying
L-theanine	Weak–Moderate	Excellent	None	Worth trying
Magnesium L-threonate	Moderate (cognition)	Good	None	Worth trying (cognition focus)
Melatonin (low-dose)	Moderate (narrow)	Good short-term; CV caution long-term	None	Use case-specific, not nightly forever
Ashwagandha	Moderate	Moderate (rare hepatic; interactions)	None	Selective use, root extract only
Glycine	Weak	Excellent	None	Optional
Apigenin	Weak/preliminary	Good	None	Optional
Phosphatidylserine	Weak/preliminary	Good	None	Optional
Valerian	Weak/mixed	Good	Low	Skip
Taurine	Not a sleep aid	Good	None	Not for sleep
GABA (oral)	Weak/negative	Good	None	Skip
CBD	Preliminary	Variable (drug interactions)	Low	Skip absent indication