Magnesium
Magnesium (various salts)
Written by Aaron CuhaReviewed Sep 2026
Also known as: Magnesium glycinate, Magnesium bisglycinate, Magnesium citrate, Magnesium oxide, Magnesium L-threonate, Magtein
The blood pressure effect is real and small, about 2 mmHg. The sleep evidence is three trials in 151 people rated low to very low certainty. The brain magnesium claim traces to rats, a retracted paper, and three trials all funded by the seller.
Overview
Magnesium is the supplement where the gap between marketing confidence and evidence quality is widest, and the useful thing about it is that the whole chain can be traced.
What holds up: a lot of people eat less magnesium than the estimated average requirement, serum magnesium is a genuinely poor status marker because the body defends it at the expense of tissue stores, and supplementation lowers blood pressure by a small, real amount across 34 randomised trials.
What does not hold up nearly as well as claimed: sleep. After decades of marketing, the randomised base for magnesium and insomnia in older adults is three trials totalling 151 people, rated low to very low certainty by the review authors themselves, who wrote that the literature is substandard for making recommendations. No trial has measured sleep objectively.
And the forms hierarchy sold to consumers, glycinate for sleep and threonate for brain, has no randomised head-to-head support on any clinical endpoint at all.
Mechanism of action
Magnesium is a cofactor for more than 300 enzymes, including every reaction using ATP, since the physiological substrate is Mg-ATP. It is a natural calcium channel antagonist, which is the most plausible route to the blood pressure effect, and it modulates NMDA receptor activity, which is the basis of the neurological claims. Body distribution explains the testing problem: roughly 53% of total body magnesium sits in bone, 27% in muscle and 19% in soft tissue, with only about 0.3% in serum, so a serum value can be normal while tissue stores are depleted.
Human evidence
Substantial randomised evidence for blood pressure and glucose, with small effect sizes. Thin and low-certainty evidence for the uses it is most heavily marketed for, namely sleep and cognition.
- Blood pressure: 34 trials, 2,028 participants, systolic minus 2.00 mmHg. In metabolically unwell populations the figure roughly doubles to 4.18 mmHg across 11 trials and 543 participants.
- Sleep: 3 trials, 151 older adults, GRADE low to very low. Sleep onset latency 17 minutes shorter, total sleep time not significant.
- Glycinate specifically, the form most marketed for sleep, has exactly one randomised trial: 155 adults with self-reported poor sleep, Cohen's d of 0.2 at p = 0.049 on a self-report scale, with an author who directs a contract research organisation funded by nutraceutical companies.
- Forms: no trial anywhere randomises people to two magnesium forms and compares them on a clinical endpoint. Every form comparison uses a blood, urine or saliva marker, in samples of at most 46 people.
- Magnesium L-threonate cognition: three human trials, all funded by the entity commercialising the branded ingredient. No independent replication exists. One of the trials was a five-ingredient formula including phosphatidylserine, so no effect can be attributed to magnesium.
- The rat work that started the brain magnesium story measured synaptic density, NMDA receptor expression and brain magnesium. No human trial has measured any of those three.
What this does not tell you: Serum magnesium can be normal in total-body depletion, which is a real limitation. It does not follow that the alternatives sold as superior are better: a systematic review of 20 candidate biomarkers found serum, red cell and urinary magnesium usable and could draw no conclusion about ionised magnesium or the loading test. The mortality and cardiovascular associations are observational and are for dietary intake, not supplements. Note also the recurring author affiliation across the blood pressure meta-analyses: the Center for Magnesium Education and Research is an advocacy organisation, which does not make the findings wrong but belongs in the reader's view.
Reading the research record
Magnesium is where this site's method earns its keep, because the marketing claim and the evidence are both traceable and they do not match. Take the threonate and brain story, which is the most aggressively sold claim in the category. It begins with a 2010 rat study from the laboratory that developed the compound. The next link in the chain, a mouse Alzheimer's paper, was retracted from the Journal of Neuroscience by its own authors and republished in another journal. The human trial that became the headline was registered with a biomarker as its primary outcome and anxiety and sleep as its subject, reported a derived four-domain cognitive composite that was not a registered outcome at all, and declined to report the anxiety and sleep results it was powered for. The sponsor that sells the compound designed the study, ran the statistical analysis for the cognitive endpoints, and wrote the paper. Two later trials were likewise funded by parties commercialising the ingredient, and one of them tested a five-ingredient formula.
None of that proves magnesium threonate does nothing. It means the claim that it raises brain magnesium and builds synapses in humans is entirely unmeasured in humans, and what carried over from the rat work to the label was a compound name and an analogy.
The evidence, charted
Fig. 1 · evidence composition
1of 7 citations (14%) are in people
Counted from the citation list on this page. The Human count is the same number shown in the badge at the top. Both understate any literature larger than the sources we cite.
Fig. 2 · evidence over time
Evidence spans 5 distinct years, 2003 to 2021, counted from the citation list on this page.
Fig. 3 · legal status at a glance
US
Not approved
UK
Not approved
AU
Not approved
CA
Not approved
Not approved in any of the four jurisdictions shown. A jurisdiction's classification is a regulatory fact, not a verdict on the science; see Legal status below for the full text and any notes.
Key studies & citations
- Review2016
Effects of magnesium supplementation on blood pressure: a meta-analysis of randomized double-blind placebo-controlled trials
34 randomised trials, 2,028 participants, median 368 mg a day for a median three months. Systolic pressure fell 2.00 mmHg (95% CI 0.43 to 3.58) and diastolic 1.78 mmHg (0.73 to 2.82). Real, small, and the authors note residual heterogeneity. A co-author is affiliated with the Center for Magnesium Education and Research, an advocacy body.
Hypertension - Review2017
The effect of magnesium supplementation on blood pressure in individuals with insulin resistance, prediabetes, or noncommunicable chronic diseases: a meta-analysis of randomized controlled trials
11 trials, 543 participants who were metabolically unwell. Mean reduction 4.18 mmHg systolic and 2.27 mmHg diastolic, roughly double the all-comers figure. Both standardised effect confidence intervals nearly touch zero, and the paper drew two published comments, meaning it was formally contested.
American Journal of Clinical Nutrition - Review2021
Oral magnesium supplementation for insomnia in older adults: a systematic review and meta-analysis
The entire randomised base for magnesium and insomnia in older adults: 3 trials, 151 participants. Sleep onset latency 17.36 minutes shorter (95% CI 27.27 to 7.44), total sleep time improved by 16 minutes and was not statistically significant. GRADE rating low to very low, with all trials at moderate to high risk of bias. The authors write that the quality is substandard for physicians to make well-informed recommendations.
BMC Complementary Medicine and Therapies - Review2014
Author-initiated retraction: Elevation of brain magnesium prevents and reverses cognitive deficits and synaptic loss in Alzheimer's disease mouse model
A key preclinical paper in the magnesium threonate and Alzheimer's story was retracted from a leading neuroscience journal by its own authors, then republished elsewhere four months later. Material for anyone weighing the brain magnesium claim.
Journal of Neuroscience - Human2003
Mg citrate found more bioavailable than other Mg preparations in a randomised, double-blind study
46 healthy people, 300 mg elemental magnesium daily as amino acid chelate, citrate or oxide for 60 days. Citrate led on serum and urinary markers. Crucially, the endpoint closest to tissue status, red cell magnesium, showed no difference between forms, and no clinical outcome was measured. This is the most-cited head to head that exists.
Magnesium Research - Review2017
Intestinal absorption and factors influencing bioavailability of magnesium: an update
The expert review position, stated plainly: the type of magnesium salt appears less relevant than is often thought, and it remains unclear which binding form gives the highest bioavailability. Dose and existing status matter more than form.
Current Nutrition and Food Science - Review2016
Dietary magnesium intake and the risk of cardiovascular disease, type 2 diabetes, and all-cause mortality: a dose-response meta-analysis of prospective cohort studies
Observational. 40 cohorts, over a million participants. Per 100 mg a day of dietary magnesium: total cardiovascular disease was null at 0.99, coronary heart disease null at 0.92, while heart failure was 0.78, stroke 0.93, type 2 diabetes 0.81 and all-cause mortality 0.90. Dietary intake, not supplements.
BMC Medicine
Frequently asked questions
Which form of magnesium should I take?
The honest answer is that nobody knows, because no trial has ever randomised people to two forms and compared them on a clinical outcome. The expert review position is that the salt matters less than commonly thought and that dose and your existing status matter more. Oxide is genuinely poorly absorbed. Beyond that, the hierarchy you have been sold rests on absorption markers in studies of at most 46 people.
Does magnesium help you sleep?
Possibly a little, and the evidence is much weaker than the marketing. Three randomised trials in 151 older adults pooled to a 17-minute reduction in time to fall asleep, rated low to very low certainty, with total sleep time not significantly changed. No trial has measured sleep objectively rather than by questionnaire. The review's own authors called the literature substandard.
Is a serum magnesium test useful?
Partly. Only about 0.3% of body magnesium is in serum and the body defends that number, so a normal result does not rule out depletion. Red cell magnesium and urinary excretion are the other two markers that a systematic review found usable. Ionised magnesium and loading tests, often sold as superior, have not been validated.
Does it lower blood pressure?
Yes, by about 2 mmHg systolic across 34 randomised trials, and by roughly double that in people who are metabolically unwell. That is a real effect and a small one. It is worth having as part of several small things, and it is not a substitute for treatment if your pressure needs treating.