Transcript
The sleep supplement market rests on a premise worth examining carefully: that a capsule can fix poor sleep. The global industry is now worth billions, and the shelves are stocked with magnesium, glycine, L-theanine, melatonin, and combinations of all four. The problem is that most people reaching for these supplements have not identified what is actually disrupting their sleep. And if the mechanism of a supplement does not match the mechanism of your problem, you are probably wasting your money — and possibly postponing an intervention that would actually help.
This episode works through two related questions: does the strongest available evidence support these four compounds as genuine sleep aids, and is melatonin in particular being systematically misused — taken as a sedative when its actual pharmacological role is something quite different?
Start with what sleep is, biochemically, because the mechanisms matter here. Sleep is not simply the absence of wakefulness. It is an actively regulated state governed by two interacting systems. The first is the circadian clock — a roughly 24-hour oscillation driven by light exposure and mediated by hormones, particularly melatonin secreted from the pineal gland. The second is homeostatic sleep pressure — the accumulation of adenosine and other sleep-promoting compounds during waking hours. A supplement that acts on one system will not necessarily help if your problem originates in the other. This distinction is central to understanding why most of these compounds are either oversold or chronically misapplied.
Begin with magnesium, because the evidence base is the most developed and also the most instructive in terms of what rigorous research actually looks like when applied to a popular supplement.
Magnesium is the fourth most abundant mineral in the body and a cofactor in over three hundred enzymatic reactions. Its relevance to sleep comes through several pathways simultaneously. It binds to GABA-A receptors — the same receptor class targeted by benzodiazepines and alcohol — and in doing so reduces neuronal excitability. It also antagonises NMDA receptors, which are excitatory glutamate receptors; blocking these reduces the kind of hyperarousal that keeps people awake despite fatigue. Separately, magnesium participates in regulating the HPA axis, meaning adequate magnesium is associated with lower cortisol output. And magnesium serves as a cofactor in the enzymatic conversion of serotonin to melatonin in the pineal gland, which gives it relevance to both the calming side of sleep and the circadian timing side.
That mechanistic breadth is genuinely interesting. But mechanisms do not automatically translate into clinically meaningful effects in well-nourished people, and this is where the evidence becomes more complicated.
The most rigorous recent trial, published in 2025, enrolled 155 adults between 18 and 65, randomised them to 250 milligrams of elemental magnesium daily in the bisglycinate form or to placebo, and tracked insomnia severity over four weeks. The magnesium group showed a statistically significant reduction in Insomnia Severity Index scores compared to placebo — a drop of about 3.9 points versus 2.3. That difference is real. But the effect size, expressed as Cohen's d, was 0.2. In clinical research, 0.2 is considered small. Both groups improved; the supplement group improved somewhat more. And notably, the participants who gained the most were those with lower dietary magnesium intake at baseline.
An earlier trial from 2012, conducted in elderly patients with insomnia, used 500 milligrams of elemental magnesium daily and found more convincing improvements — better sleep efficiency, shorter onset latency, longer total sleep time — confirmed by actigraphy rather than just self-report. Elderly populations are more likely to be genuinely magnesium-insufficient, because absorption declines with age and dietary intake often narrows. That likely explains why the effect was larger there.
Then look at the large-scale observational data. The CARDIA study, which followed thousands of adults over decades and assessed dietary magnesium intake against sleep duration, found no statistically significant association between magnesium intake quartiles and short sleep duration in their fully adjusted models. The confidence intervals crossed one broadly. Whatever signal exists at the population level is weak enough that it disappears when you account for the other things that correlate with eating a magnesium-rich diet — fruit, vegetables, whole grains, stable income, consistent eating patterns.
There is also an important wrinkle around form. Magnesium oxide is the cheapest and most commonly sold form, but it is poorly absorbed in the gut and functions primarily as a laxative at higher doses. Better-absorbed forms — glycinate, bisglycinate, threonate — should theoretically outperform oxide on sleep outcomes by getting more magnesium into circulation. But one review noted that magnesium chloride showed no significant sleep improvement while magnesium oxide actually showed benefit at lower doses in one study, which cuts against the simple bioavailability hypothesis. The research here is genuinely messy.
Magnesium threonate — a form developed specifically to cross the blood-brain barrier — showed improvements in deep sleep and REM stages over six weeks in a 2022 study using 1,500 milligrams of the compound, which delivers roughly 144 milligrams of elemental magnesium. The blood-brain-barrier penetration is pharmacologically plausible, but this is a single study and the effect sizes have not been independently replicated in large trials.
What the magnesium evidence actually says is this: if you have documented insufficiency, are elderly, or have consistently low dietary intake, supplementation probably helps, and the bisglycinate or threonate forms are better choices than oxide. If you are a well-nourished adult with normal magnesium status, the effect size in the best available trial is small, and the large observational studies find no population-level signal. The recommended daily intake is 310 to 360 milligrams for women and 400 to 420 milligrams for men, and the tolerable upper limit for supplemental magnesium is 350 milligrams of elemental magnesium per day — above which gastrointestinal effects become likely.
One important caveat on measuring status: serum magnesium testing, which is the most commonly used clinical measure, may not reliably identify who will respond to supplementation, because the body maintains serum levels at the expense of intracellular and bone stores. Normal serum magnesium does not rule out functional insufficiency in the tissues that matter. The 2025 trial explicitly called for longer interventions and objective sleep measurement — polysomnography or actigraphy — in future research. Those trials do not yet exist at scale.
Now glycine. The evidence here is thinner, but the mechanism is genuinely interesting. Glycine is the simplest amino acid and also an inhibitory neurotransmitter in the spinal cord and brainstem. The sleep-relevant mechanism appears to operate peripherally: glycine causes mild vasodilation, which increases heat dissipation from the skin. Core body temperature needs to drop to initiate and maintain sleep, and anything that accelerates that drop should theoretically advance sleep onset and improve sleep quality.
The human trial evidence is small but consistent. Studies — mostly from Japanese research groups — have used doses of around three grams taken before bed and found reductions in sleep onset latency, improvements in subjective sleep quality, and reduced daytime fatigue the following day. The participants in these studies were not clinically insomniac; they were people with self-reported poor sleep. Effect sizes are modest. The mechanistic story involving body temperature is biologically coherent and fits with what is known about sleep thermoregulation.
The problem is that these trials are small, many come from a single research group, and they rely heavily on self-report rather than polysomnography. There is no large independent replication. Glycine is safe and cheap, which means it does not attract the pharmaceutical funding that would generate large randomised controlled trials. The evidence gap here is partly a market failure, not just a scientific one.
Magnesium bisglycinate, notably, combines magnesium with glycine as the chelating agent. This raises a still-unresolved question: in trials of magnesium bisglycinate, how much of the effect is from magnesium and how much from glycine independently? The 2025 bisglycinate trial did not include a glycine-only arm, so it cannot separate those contributions. That is a real limitation in the current evidence base.
L-theanine is an amino acid found almost exclusively in tea leaves — predominantly green and white tea. It does not behave like a classical sedative. Instead, it promotes what researchers describe as relaxed alertness: it increases alpha-wave activity in the brain, associated with calm focus rather than drowsiness. Its primary sleep-relevant effect appears to be on anxiety-related sleep disruption rather than on sleep architecture itself.
Evidence from randomised trials suggests L-theanine reduces the time it takes to fall asleep and may improve sleep efficiency, particularly in people whose sleep problems are driven by rumination, worry, or elevated sympathetic arousal. It appears to modulate glutamate activity and increase GABA, which would explain the calming effect. Some trials have also found benefits in children with ADHD, where sleep disruption is often driven by anxiety and hyperarousal.
The treatment-mismatch problem is worth making explicit here. If your sleep difficulty is primarily circadian — shifted sleep phase, jet lag, irregular schedule — L-theanine is unlikely to help much. If your difficulty is primarily homeostatic — accumulated sleep debt, irregular sleep timing — it also does not address the root cause. Where it fits is in the narrower category of hyperarousal: the person who is tired but cannot switch off. For that specific presentation, there is reasonable preliminary evidence. But trial quality is mixed, sample sizes are small, and the effect does not generalise across all sleep complaints.
Now melatonin — and this is where the misuse question is most stark.
Melatonin is not a sedative. It does not directly induce sleep. What it does is signal darkness to the brain's circadian clock, specifically the suprachiasmatic nucleus in the hypothalamus. Melatonin secretion rises in the evening as light fades, and this rise is part of the biological signal that shifts the body toward sleep phase. Taking exogenous melatonin does not knock you out. What it does, if timed correctly, is shift the phase of your circadian clock.
The evidence for melatonin is actually quite strong — but only for specific, phase-related problems. Jet lag is the clearest case: taking melatonin at the destination's bedtime helps the circadian clock re-synchronise faster. Timing matters enormously here; melatonin taken at the wrong circadian phase can shift the clock in the wrong direction. Shift workers and people with delayed sleep phase syndrome — where the body's clock runs several hours late — are also genuine candidates for appropriately timed melatonin.
The problem is that most people taking melatonin are not using it as a circadian tool. They are taking it as a sedative at high doses — often three to ten milligrams — when the physiological dose needed to produce the circadian signalling effect is closer to 0.5 milligrams. The trials that demonstrate phase-shifting effects have typically used low doses with careful attention to timing relative to the individual's existing circadian phase. High-dose, timing-agnostic use — which characterises most consumer behaviour — is not well supported by the mechanistic evidence, and at those doses melatonin levels may rise well above the normal physiological range for unclear benefit.
There is also a habituation question worth acknowledging. Melatonin appears to have low abuse potential, but whether long-term exogenous supplementation affects endogenous melatonin production is not fully resolved. The concern is theoretical rather than established, and the long-term pharmacology of nightly high-dose melatonin in healthy adults has not been rigorously studied.
So where does all of this land?
The claim that most sleep supplements treat the wrong problem holds up reasonably well against the evidence — with precision required about what "wrong problem" means in each case. Magnesium has a plausible mechanism and real trial evidence, but the effect is modest and most pronounced in people who are actually insufficient. It acts by reducing neuronal excitability and possibly supporting endogenous melatonin synthesis — it is not acting on sleep pressure or circadian timing directly. If your sleep problem is driven by genuine magnesium insufficiency, there is something to work with. If you have adequate dietary magnesium and normal sleep architecture, the evidence for supplementation is weak.
Glycine shows promise in small trials through a thermoregulatory mechanism, but independent replication is lacking. L-theanine may help with anxiety-driven sleep disruption, and that is a real and common presentation — but it does not address circadian or homeostatic dysregulation. Melatonin is genuinely effective for phase-related problems and genuinely oversold as a general sleep aid. The doses people commonly take are far higher than what the circadian evidence supports, and the timing most people use is not calibrated to their individual circadian phase.
None of these compounds repairs disrupted sleep hygiene, irregular schedules, excessive light exposure at night, or the chronic hyperarousal that follows months of poor sleep. They work, when they work, through distinct mechanisms, and none of those mechanisms is broad enough to cover uncharacterised sleep complaints.
Several questions remain genuinely open. Whether serum magnesium testing can reliably identify who will respond to supplementation is not established. The independent contribution of glycine within bisglycinate formulations has not been separated in any published trial. The optimal dose and timing for melatonin across different circadian phenotypes is an active research question. And the long-term effects of regular supplementation with any of these compounds — on endogenous systems, on sleep architecture, on whether supplementation eventually becomes self-defeating — are areas where the evidence does not yet exist at the necessary scale or duration.