Does magnesium actually help you sleep?

Most of the supplements I write about here are exotic. Magnesium is not. It is an essential mineral, a genuine share of the population falls short of the recommended intake, and the mechanism story involves nothing more mystical than a nutrient your nerves and muscles actually require.

That is exactly why it needs reading carefully. When a claim sounds reasonable, plausibility quietly does the work that evidence is supposed to do, and nobody goes looking for the trial. So let us go look at the trial.

The trial worth reading

In August 2025, Julius Schuster, Igor Cycelskij, Adrian Lopresti and Andreas Hahn published a randomized, double-blind, placebo-controlled trial of magnesium bisglycinate in Nature and Science of Sleep. It is a good study by supplement-industry standards, which is the reason it is worth taking seriously rather than dismissing.

The design: 155 adults randomized, aged 18 to 65, all reporting poor sleep for more than four weeks and scoring above 12 on the Regensburg Insomnia Scale. People with diagnosed sleep disorders, shift workers, and anyone on sleep-affecting medication were excluded. Four weeks of capsules taken 30 to 60 minutes before bed, against a visually identical cellulose placebo, with randomization stratified by insomnia score, sex and age. It was funded by the Institute of Food and One Health at Leibniz University Hannover as an internal academic project, not by a supplement brand, which is worth more than it sounds. One author, Adrian Lopresti, discloses that he is "the managing director of Clinical Research Australia, a contract research organization that receives research funding from nutraceutical companies."

The primary outcome was the Insomnia Severity Index. Here is what it found. The magnesium group improved by 3.9 points, the placebo group by 2.3 points, for a between-group difference of 1.6 points, 95% confidence interval 0.0 to 3.3, p = 0.049. The effect size was Cohen's d = 0.20, which the authors correctly label small.

Sit with that confidence interval for a second. The lower bound is zero. The p-value is 0.049, not 0.0049, meaning it clears the conventional threshold by a hair. This is the weakest shape a positive result can have while still being called positive, and a headline that says "magnesium improves sleep" is technically reporting it accurately.

Then look at everything else they measured

The secondary endpoints are where a trial tells you whether it found a real effect or a lucky one, because a genuine improvement in sleep should show up in more than one instrument. Every single secondary outcome came back statistically null: the Regensburg Insomnia Scale (p = 0.186), a sleep quality scale (p = 0.069), the Epworth Sleepiness Scale (p = 0.624), fatigue severity (p = 0.391), perceived stress (p = 0.708), and a depression and anxiety screen (p = 0.634).

The sleep diary, which would have given day-by-day detail, had a completion rate under 10% and the authors report it as uninterpretable. And all of it was self-reported. Nobody was wired up, nobody was measured while asleep. The authors say so plainly: "The validity of the sleep-related outcomes is limited by the exclusive reliance on self-reported data without objective verification."

They are also honest about how much the pill accomplished for the people it did help: "While magnesium bisglycinate significantly improved ISI scores, the mean score at week 4 remained in the subthreshold insomnia range, indicating that supplementation alone is unlikely to eliminate insomnia in many individuals."

What was in the capsule

Now the part that is my actual job, and the part no coverage of this trial mentioned.

Read the methods and you find that each capsule contained 893 mg of magnesium bisglycinate, which breaks down to 125 mg of magnesium and 761.5 mg of glycine. Two capsules a day. So the daily dose was 250 mg of magnesium and 1,523 mg of glycine.

Glycine is not a passenger. It is an amino acid with its own sleep literature, and the paper says so itself: glycine "has also been explored for its sleep-promoting properties due to its ability to interact with key neurotransmitter systems, including the N-methyl-D-aspartate (NMDA) receptors," noting that "some studies suggest that glycine supplementation at a dose of 3 g can enhance sleep quality and reduce daytime fatigue." The authors even write that its "co-presence may produce synergistic effects with magnesium in supporting sleep physiology."

There was no glycine-only arm. There was no magnesium-in-another-form arm. Which means this trial, whatever it detected, cannot tell you which of the two ingredients did it, or whether it took both.

And I want to point at one detail in the exclusion criteria, because it settles the question of whether the researchers considered glycine active: they excluded anyone who had taken more than 1 g of glycine per day in the previous month. You do not control for an inert substance. They knew.

This is the recurring structural problem in my industry and I have written about it with a six-mushroom endurance blend and a five-mushroom fatigue study. The unit of evidence is the formula that was tested, not the ingredient on the front of the box. When a two-ingredient product is studied as though it were one ingredient, the marquee name inherits credit it did not earn on its own.

The measurement that never happened

Here is the omission that bothers me most, and it is the one that would have made this study genuinely decisive.

Nobody's magnesium was measured. Not serum, not urinary excretion, not at baseline, not at the end. In a trial about a nutrient.

The authors did run an exploratory analysis, and it points somewhere interesting: participants who reported eating fewer magnesium-rich foods improved more (Spearman's rho = −0.25, p = 0.036), with no such correlation in the placebo group. That is the repletion hypothesis, the idea that magnesium helps people who are short of magnesium and does little for people who are not.

But the dietary intake was captured on an unvalidated 1-to-5 scale, and the correlation was only run in the treatment arm. The authors flag this against themselves: "the use of an unvalidated questionnaire to assess dietary intake and the absence of biochemical markers of magnesium status raise the possibility that factors such as overall dietary quality or health status may have influenced the results." Their own recommendation for next time is serum magnesium or, better, 24-hour urinary excretion.

For a nutrient, repletion versus pharmacology is the entire question. Are you fixing a shortfall, or is this compound doing something drug-like in people who already have enough? You answer that by measuring status. A nutrient trial with no status measurement is a drug trial that never recorded the dose, and it is a strange thing to leave out of an otherwise carefully run study.

What happens at population scale

Six months later, in February 2026, a very different kind of study landed in the European Journal of Nutrition. Noelle Stadie, Raphaël Heinzer and Pedro Marques-Vidal went to the CoLaus|PsyCoLaus cohort in Switzerland and looked at magnesium supplement users across three follow-ups: 3,887 people in 2009-12, 1,916 in 2014-17, and 1,561 in 2018-21. Crucially, they had both questionnaires and polysomnography, the overnight sleep-lab measurement the randomized trial lacked entirely. They also built a polygenic risk score from five variants related to magnesium levels.

Their conclusion, verbatim: "We found no consistent association between magnesium supplementation and subjective or objective sleep parameters or restless leg syndrome. Magnesium supplementation did not prevent night cramps."

They found something that looks worse than nothing, too: magnesium users had a higher likelihood of night cramps and restless legs, in both the cross-sectional and the prospective analyses.

Now, please do not read that as magnesium causing cramps, and I would be doing exactly what I criticize if I let you. This is confounding by indication, one of the most common traps in nutrition epidemiology. People start taking magnesium because they get cramps. An observational study of supplement users is, before anything else, a study of who chose to take the supplement, and the reason they chose it walks into the data with them.

So this study cannot prove magnesium does not work, and it does not claim to. What it contributes is the thing the trial could not: objective, instrumented sleep measurement at population scale, in which no signal appeared.

Where that leaves you

Putting both honestly side by side: a small, borderline, entirely self-reported benefit in poor sleepers over four weeks, from a capsule that also delivered 1.5 g of glycine, with every secondary measure null and no one's magnesium ever measured. And at population scale, with real sleep labs, nothing.

That is not a debunking, and I am not going to inflate it into one. Magnesium is a genuine nutrient with a genuine deficiency state, and the defensible version of the claim survives all of the above: if your intake is low, correcting it is worth doing on its own merits, and any sleep benefit is most likely a repletion effect rather than a sedative one. Food first, since dietary magnesium comes attached to the rest of a decent diet. The safety profile in the trial was reassuring, with 2 adverse events in the magnesium group against 7 on placebo and nothing serious.

Low risk is not the same as demonstrated benefit, though, and the gap between those two is where a great deal of money gets spent. If you are sleeping badly enough that it matters, the evidence for behavioral treatment of insomnia is in a different league from anything above, and a 1.6-point questionnaire shift is not a reason to delay it.

Not medical advice. This is educational analysis, not a recommendation — a study is not a prescription. Talk to a qualified clinician before acting on anything you read here. Full disclaimer →

The signal

The most important number in a supplement study is often one that was never collected. Here it was a blood test that would have cost a fraction of the trial budget and would have told you whether you were watching a deficiency get corrected or a molecule do something. Without it, a genuinely careful study leaves its central question open.

The reading skill this hands you generalizes past magnesium. When you see a supplement result, ask three things: how many ingredients were actually in the capsule, did anything other than a questionnaire get measured, and did they measure the thing the product is supposed to be fixing. I applied the same three questions to a study a regulator took apart and to an effect size that was too big to be true. They keep working, because the ways a study can quietly fall short of its own headline are surprisingly few, and they repeat.

Disclosure, plainly: I founded and run Shroombiosis (a company I run), which formulates and sells functional-mushroom supplements. That is a direct commercial stake in the industry this report is about, and the standard I am applying to this trial is the standard my own labels have to meet. Shroombiosis does not sell a magnesium product, and nothing here is sponsored or earns a commission; here's the full policy. A recommendation with no stake at all: for performance nutrition, Die Tryin Co. is a fellow combat-veteran-owned brand I'm glad to point people to. I don't own it and earn nothing from the link.