Supplements for sleep: what the evidence supports

Sleep research does not measure one thing, so we grade separate sleep outcomes. A supplement can shift one measure, such as the time it takes to fall asleep, and leave the others unchanged.

Table: Sleep outcomes we grade and how studies measure them

The Pittsburgh Sleep Quality Index (PSQI) is a self-rated questionnaire covering the previous month: 19 items produce seven component scores and a global score [1]. In its validation study, which compared healthy good sleepers with poor sleepers from patient groups, a global score above 5 separated the two groups with 89.6% sensitivity and 86.5% specificity [1].

Questionnaires and devices do not always agree. In exercise trials, self-rated sleep quality improved while actigraphy, a wrist-worn movement sensor, found few changes apart from sleep efficiency [3]. The authors of one supplement trial note that the consumer wrist tracker they used tends to overestimate total sleep time and sleep efficiency compared with polysomnography, the laboratory sleep recording [4].

The habits below have their own trial evidence, and none of them involves a supplement.

Regular exercise. In a 2021 meta-analysis of 22 randomized controlled trials in adults without diagnosed disease, exercise programs lasting at least 2 months improved self-rated sleep quality compared with control (PSQI mean difference -2.19) [3]. A 2019 meta-analysis of 23 studies comparing a single evening exercise session with no exercise in healthy adults concluded that the studies did not support the idea that evening exercise worsens sleep, "in fact rather the opposite", though vigorous exercise ending 1 hour or less before bedtime might delay sleep and shorten it [5].

Caffeine timing. In a 2023 meta-analysis, caffeine reduced total sleep time by 45 minutes and lengthened the time to fall asleep by 9 minutes [6]. To avoid a loss of total sleep time, the authors estimated a cut-off of 8.8 hours before bedtime for coffee (107 mg per 250 mL) [6]. In a placebo-controlled study of sleep at home, 400 mg of caffeine taken 6 hours before bedtime still disrupted sleep [7].

Alcohol. In a 2025 meta-analysis of 27 studies in healthy adults, about two standard drinks were enough to disrupt REM (rapid eye movement) sleep, and the disruption grew with larger amounts [8]. Falling asleep faster was seen only after high doses of about five standard drinks, and the effect on total sleep time was too uncertain to call [8].

A regular schedule. A 2025 systematic review of 59 observational studies linked irregular sleep timing with poorer mental, metabolic, vascular, cognitive and survival outcomes [9]. These are associations from cohort data, not trial results, and the authors call for randomized trials of schedule changes [9].

When poor sleep persists. CBT-I is a structured cognitive behavioral program that combines at least three of these parts: cognitive therapy, stimulus control, sleep restriction, sleep hygiene and relaxation [2]. In a meta-analysis of 20 trials with 1,162 adults with chronic insomnia, a clinical population, CBT-I shortened the time to fall asleep by 19.03 minutes and cut time awake during the night by 26.00 minutes, compared with inactive comparators [2]. The gain in total sleep time, 7.61 minutes, had a confidence interval that included zero, and no adverse outcomes were reported [2]. Results in clinical groups may not apply to healthy adults; a clinician can say whether this approach fits.

The ranked list on this page is ordered by each compound's approved grades in our evidence matrix; the summaries below take the studies compound by compound. The rubric is on how we grade evidence. Safety, interactions and legal status for each compound are on its monograph; anyone taking medicines should check with a doctor or pharmacist before adding a supplement.

Melatonin

Melatonin is a hormone the brain produces in response to darkness [35]. In a 2005 meta-analysis of 17 studies (284 participants), 15 of which enrolled healthy people or people whose only condition was insomnia, melatonin shortened the time to fall asleep by 4.0 minutes and added 12.8 minutes of total sleep compared with placebo. The authors describe the included studies as highly heterogeneous [20]. For sleep onset and total sleep, results are mixed: a 2022 meta-analysis of 24 trials in people with chronic insomnia, a clinical population, found no significant effect in adults on time to fall asleep, total sleep time or sleep efficiency [22]. A 2022 meta-analysis of 23 trials in adults with various diseases found a 1.24-point PSQI improvement, with high heterogeneity and no effect in some disease groups [25]. A Cochrane review found that 9 of 10 placebo-controlled trials in air travelers, airline staff and military personnel reported less jet lag when melatonin was taken close to the target bedtime after flights crossing five or more time zones [21].

Legal status differs by country (see the FAQ). Canada's licensing terms, doses studied, label accuracy and safety are on the melatonin monograph.

Ashwagandha

In a 2021 meta-analysis of five randomized controlled trials with 400 participants, ashwagandha extract had a small effect on overall sleep compared with placebo (standardized mean difference -0.59) [10]. The effect was more prominent in adults diagnosed with insomnia, a clinical subgroup, and in trials using 600 mg/day or more for 8 weeks or longer [10]. All five trials were run in India, and one was partly sponsored by the manufacturer [10]. See the ashwagandha monograph for safety and interactions.

Saffron

A 2022 meta-analysis of randomized trials found saffron lowered PSQI scores by 2.14 points compared with placebo, pooling healthy adults with clinical groups [23]. Its literature search ran to October 2021 [23]. In a later 4-week trial in 165 adults reporting moderate insomnia, supported by the extract maker Activ'Inside, saffron improved Athens Insomnia Scale scores more than placebo, and all other measures showed no significant differences [27]. Other trials and their funding are on the saffron monograph.

Magnesium

In a 2021 meta-analysis of three placebo-controlled trials in 151 older adults with insomnia, a clinical population, magnesium shortened the time to fall asleep by 17.36 minutes, but all trials were at moderate-to-high risk of bias and the evidence was of low to very low quality [17]. In a randomized, double-blind trial in 155 adults reporting poor sleep, funded by Leibniz University Hannover (one author runs a contract research organization funded by nutraceutical companies), 250 mg/day of elemental magnesium as bisglycinate lowered Insomnia Severity Index scores more than placebo by week 4, a small effect (Cohen's d = 0.2) [18]. A 2026 systematic review of 12 trials found results inconsistent across questionnaires, EEG and actigraphy, and rated certainty low to very low [19]. More on the magnesium monograph.

Glycine

Glycine is a conditionally essential amino acid [14]. A 2026 scoping review, searched through July 9, 2026, found three trials of glycine and sleep, all reporting a benefit on at least one sleep-related outcome, and called for larger trials with standardized sleep outcomes [14]. A 2024 systematic review judged the glycine sleep studies in healthy people small and at high risk of bias [26]. In a single-blinded crossover trial run by researchers employed by Ajinomoto, ten healthy men slept 25% less than usual for three nights, and 3 g of glycine before bed reduced next-day fatigue ratings compared with placebo, with seven men analyzed [15]. The glycine monograph covers the sleep-onset trial and each trial's company links.

L-theanine

A 2025 meta-analysis of 18 articles (897 participants across the 19 reviewed), covering all ages and health status, found an improvement in overall self-rated sleep quality with L-theanine (SMD 0.43 from 12 studies, with a wide confidence interval of 0.04 to 0.83) [16]. The authors note the lack of studies on pure L-theanine [16]. See the L-theanine monograph.

Chamomile and apigenin

The sleep trials behind this grade tested chamomile, not isolated apigenin [12][13]; in one trial, 90 mg of the chamomile extract contained 3.9 mg of apigenin [13]. A 2024 systematic review of 10 studies (772 participants) in healthy adults and adults with medical conditions pooled five of them and found chamomile lowered PSQI scores by 1.88 points, with very high heterogeneity (I2 of 88.4%); it found no improvement in sleep duration or sleep efficiency [12]. In a randomized, double-blind pilot trial in 34 adults with chronic primary insomnia, a clinical population, 270 mg of chamomile extract twice daily for 28 days showed no statistically significant difference from placebo on any sleep diary measure, though the effect size for time to fall asleep favored chamomile [13]. The study tablets were supplied by the manufacturer, MediHerb [13]. See the apigenin monograph.

Holy basil

In an 8-week randomized, double-blind, placebo-controlled trial in 100 adults with ongoing stress and current sleep problems, funded by Natural Remedies, the extract maker, 125 mg of extract twice daily improved Athens Insomnia Scale scores, a secondary outcome, more than placebo (p = 0.025) [4]. Tracker-measured total sleep time and time in bed showed no statistically significant differences between groups, and the authors say trials with gold-standard objective sleep measures are needed to confirm the sleep findings [4]. See the holy basil monograph.

5-HTP

In a single-blinded 12-week trial, 30 older adults were randomized to 100 mg of 5-HTP daily or to nothing, with no placebo [28]. The authors report a favorable effect on certain sleep quality components, and the global sleep score improved in poor sleepers but not in good sleepers [28]. See the 5-HTP monograph.

GABA

A 2020 systematic review of 14 placebo-controlled trials of oral GABA, whose first author was employed by Unilever, found very limited evidence for sleep benefits [29]. In a double-blind trial in 40 adults with insomnia symptoms, a clinical population, funded by the supplier Natural Way, 300 mg daily for 4 weeks shortened time to fall asleep more than placebo, the only measure where the groups changed differently; the placebo group had only 10 people and the groups differed at baseline [30]. See the GABA monograph.

Lemon balm

In a crossover trial with 2-week periods in 30 adults who woke unrefreshed, run by researchers employed by or consulting for the product maker Pharmextracta, Insomnia Severity Index scores were lower with lemon balm than with placebo (6.8 versus 9.7) [31]. In a 12-week double-blind trial in 60 adults with type 2 diabetes and symptoms of depression, a clinical population, sleep quality did not differ significantly from placebo [32]. See the lemon balm monograph.

NAD+ precursors

In a 12-week placebo-controlled trial in 60 older adults, run by researchers employed by Meiji Holdings, 250 mg/day of NMN (nicotinamide mononucleotide) lowered global PSQI scores compared with placebo [33]. A 2026 network meta-analysis of 27 studies reported that no precursor significantly improved sleep quality, though sleep quality was left out of its network analysis for lack of connectivity [34]. See the NAD+ precursors monograph.

Uridine

The sleep findings for uridine come from animal work: a 2001 review describes uridine derivatives injected into the brains of mice, and sleep studies in rats, and notes that the structure of the proposed uridine receptor had not yet been worked out [24]. As of November 11, 2026 we found no controlled trials of uridine on sleep in adults (PubMed and Europe PMC search, November 11, 2026). See the uridine monograph.

Caffeine: evidence of harm

Caffeine is on this page because of what it does to sleep, and its grade sits outside the ranking. In a 2025 meta-analysis of 22 controlled crossover trials with 956 healthy adults, caffeine reduced total sleep time by 34.67 minutes and lengthened the time to fall asleep by 8.35 minutes [11]. Both high and low doses reduced total sleep time and sleep efficiency [11]. See the caffeine monograph for its other graded outcomes.

Partner mention, not a review. NootropicsWork earns a commission on Qualia sales. We have not reviewed Qualia Night yet, so this page makes no claims about it and links to no product. When a review appears, each of its ingredients will be judged against the same grades shown above, and a commercial relationship never changes a grade.

• Caffeine half-life: how long caffeine stays in the body, and what the timing studies found.
• The evidence matrix shows every graded sleep outcome side by side; filter it to the sleep goal.
• How we grade evidence explains each grade and its modifiers.

• A 2026 systematic review of magnesium for sleep, searched to June 1, 2026, rated the certainty of the evidence low to very low [19].
• A 2026 scoping review of glycine and collagen supplements, searched through July 9, 2026, called for larger trials with standardized sleep outcomes [14].

Grade changes from new studies are listed on what's new.

What supplements have the best evidence for sleep?

We grade each compound for each sleep outcome rather than naming one winner. Melatonin shortened the time to fall asleep by 4.0 minutes across 17 highly heterogeneous studies [20], but a 2022 meta-analysis found no significant effect in adults with chronic insomnia [22].

Is melatonin sold over the counter?

It depends on the country: it is a dietary supplement in the US [35], the NHS lists it as prescription-only in the UK [36], and Australia's Poisons Standard places it in Schedule 3 for certain products and Schedule 4 for all other human use [37]. Canada's licensing terms are on the melatonin monograph.

How long before bed does caffeine affect sleep?

The authors of a 2023 meta-analysis estimated a cut-off of 8.8 hours before bedtime for coffee (107 mg per 250 mL) to avoid a loss of total sleep time [6].

Does alcohol help you fall asleep?

Not in a useful way. Falling asleep faster was seen only after about five standard drinks, while about two were enough to disrupt REM sleep, and the effect on total sleep time was too uncertain to call [8].

Does exercising in the evening hurt sleep?

Not on average: a 2019 meta-analysis of 23 studies in healthy adults did not support that idea, though vigorous exercise ending 1 hour or less before bedtime might delay sleep [5].