Reviewed
Reviewed on 2026-09-19 against the evidence rules: every claim carries a grade, every claim and every safety flag carries a source, and every citation was re-resolved against the record it names. The read was delegated by the owner on 2026-09-18 and signed by the Lead on his behalf (D-295) — it is a delegated review, not the owner's own reading of the page. Every claim on it carries an evidence grade and a named weakest link. It is educational and is not medical advice.
Last reviewed
Two small trials of magnesium L-threonate, both funded by companies that sell it, both with real nulls inside them. For glycinate, citrate and oxide the cognitive literature is not weak — it is missing.
Safety
Adverse effects
Magnesium salts draw water into the gut, which is why several of them are sold as laxatives. Loose stools and abdominal cramping are the common effect and they scale with dose and with the form's solubility. The two trials above report the preparation as well tolerated at their doses.
Who should not
Impaired kidney function is the situation where supplemental magnesium accumulates rather than passing through, and it is a reason to ask a clinician before taking any form.
Product quality
Both cognitive trials tested magnesium L-threonate specifically, and both were funded by a company selling it — one by the preparation's developer, one by Threotech Inc, which also provided the product and helped design the study. Nobody independent has replicated either.
No data in your group
Between them the two trials cover adults 18–45 with poor sleep and adults 50–70 with cognitive complaints. A healthy adult who sleeps well is in neither group.
What the evidence says, and how strong it is
A grade belongs to one claim, in one population, for one preparation — never to a compound, a food or a page. Open a grade to see the four links between the evidence and the thing you would actually do, and which of them is doing the least work.
In a trial of 44 adults aged 50 to 70 with cognitive impairment — 23 treated, 21 on placebo — a composite cognitive score improved against placebo over 12 weeks (p = 0.003). The same trial could not determine any effect on sleep or anxiety because the placebo effects on those measures were too strong.
Evidence grade C: Early research suggestsIn whom: Adults 50–70 with cognitive impairment
What form: Magnesium L-threonate (MMFS-01), 12 weeks
Measured as: A composite of tests across four cognitive domains
Why this grade
- Weakest link
- Study quality — design, size, blinding or replication. Forty-four people at one site, with the lead author affiliated to the preparation's developer, and nobody independent has replicated it.
- What was actually studied
- Adults 50–70 with cognitive impairment (n = 44)
- Detail worth knowing
- MMFS-01 is magnesium L-threonate. 23 treated, 21 placebo, 12 weeks; composite cognitive score p = 0.003, Cohen's d = 0.91. The authors record that the effects on sleep and anxiety could not be determined because of strong placebo effects.
In a trial of 100 adults aged 18 to 45 with poor self-reported sleep, six weeks of 2 g a day improved the NIH Toolbox total cognition composite (p = 0.043) and reaction time (p = 0.031) against placebo. The trial was funded by the company that supplies the ingredient, which also helped design it.
Evidence grade C: Early research suggestsIn whom: Adults 18–45 with self-reported dissatisfied sleep
What form: Magnesium L-threonate (Magtein), 2 g/day for 6 weeks
Measured as: NIH Toolbox total cognition composite and reaction time
Why this grade
- Weakest link
- Study quality — design, size, blinding or replication. One hundred participants in a trial the ingredient's supplier funded and helped design, with no independent replication.
- What was actually studied
- Adults 18–45 with self-reported dissatisfied sleep (n = 100)
- Detail worth knowing
- 2 g/day for 6 weeks. NIH Toolbox total cognition composite p = 0.043 and reaction time p = 0.031; Raven's Progressive Matrices p = 0.953 — no difference. Self-reported sleep-related impairment improved (p = 0.043) while sleep disturbance (p = 0.316) and restorative sleep (p = 0.439) did not, and the sleep-tracking ring found no group differences in sleep. Funded by Threotech Inc, which also provided the product and helped design the study.
In the same trial, Raven's Progressive Matrices showed no difference between the groups at all (p = 0.953). The one test in that battery that looks most like reasoning is the one that did not move.
Evidence grade C: Early research suggestsIn whom: Adults 18–45 with self-reported dissatisfied sleep
What form: Magnesium L-threonate (Magtein), 2 g/day for 6 weeks
Measured as: Reasoning on Raven's Progressive Matrices
Evidence does not support
Why this grade
- Weakest link
- Study quality — design, size, blinding or replication. It is the same funder-designed trial as the claim above — which makes the null it publishes more credible than its positives, and still leaves one trial.
- What was actually studied
- Adults 18–45 with self-reported dissatisfied sleep (n = 100)
- Detail worth knowing
- 2 g/day for 6 weeks. NIH Toolbox total cognition composite p = 0.043 and reaction time p = 0.031; Raven's Progressive Matrices p = 0.953 — no difference. Self-reported sleep-related impairment improved (p = 0.043) while sleep disturbance (p = 0.316) and restorative sleep (p = 0.439) did not, and the sleep-tracking ring found no group differences in sleep. Funded by Threotech Inc, which also provided the product and helped design the study.
Self-reported sleep-related impairment improved (p = 0.043) while self-reported sleep disturbance (p = 0.316) and restorative sleep (p = 0.439) did not, and the sleep-tracking ring found no group differences in sleep at all. Resting heart rate fell and heart-rate variability rose.
Evidence grade C: Early research suggestsIn whom: Adults 18–45 with self-reported dissatisfied sleep
What form: Magnesium L-threonate (Magtein), 2 g/day for 6 weeks
Measured as: Sleep measured by a sleep-tracking ring, against sleep measured by questionnaire
Evidence does not support
Why this grade
- Weakest link
- Study quality — design, size, blinding or replication. One funder-designed trial again, and the most useful thing about it is that its objective sleep measure disagreed with its questionnaire.
- What was actually studied
- Adults 18–45 with self-reported dissatisfied sleep (n = 100)
- Detail worth knowing
- 2 g/day for 6 weeks. NIH Toolbox total cognition composite p = 0.043 and reaction time p = 0.031; Raven's Progressive Matrices p = 0.953 — no difference. Self-reported sleep-related impairment improved (p = 0.043) while sleep disturbance (p = 0.316) and restorative sleep (p = 0.439) did not, and the sleep-tracking ring found no group differences in sleep. Funded by Threotech Inc, which also provided the product and helped design the study.
No randomised trial of any of these forms against a cognitive outcome was found when this dossier was written on 2026-08-23. They are not weakly supported for cognition; they are untested for it, and the trials above used a different molecule.
Evidence grade D: Human evidence insufficientIn whom: Adults of any age
What form: Magnesium glycinate, citrate, oxide, chloride or malate
Measured as: Any cognitive outcome
Why this grade
- Weakest link
- Preparation — what was tested is not what is sold. Glycinate, citrate, oxide, chloride and malate are different molecules from the L-threonate the two trials used, and none of them has been tested against a cognitive outcome, searched on 2026-08-23.
- What was actually studied
- Adults 50–70 with cognitive impairment (n = 44); Adults 18–45 with self-reported dissatisfied sleep (n = 100)
- Detail worth knowing
- MMFS-01 is magnesium L-threonate. 23 treated, 21 placebo, 12 weeks; composite cognitive score p = 0.003, Cohen's d = 0.91. The authors record that the effects on sleep and anxiety could not be determined because of strong placebo effects.
- 2 g/day for 6 weeks. NIH Toolbox total cognition composite p = 0.043 and reaction time p = 0.031; Raven's Progressive Matrices p = 0.953 — no difference. Self-reported sleep-related impairment improved (p = 0.043) while sleep disturbance (p = 0.316) and restorative sleep (p = 0.439) did not, and the sleep-tracking ring found no group differences in sleep. Funded by Threotech Inc, which also provided the product and helped design the study.
The Cochrane review of vitamin and mineral supplementation for maintaining cognition found no strategy with a meaningful effect on cognitive decline in cognitively healthy adults in mid or late life, with moderate-certainty evidence of little or no effect from the mineral trials it could pool.
Evidence grade D: Human evidence insufficientIn whom: Cognitively healthy adults aged 40 and over
What form: Mineral supplementation, in the trials Cochrane was able to pool
Measured as: Overall cognitive function and incidence of cognitive impairment
Evidence does not support
Why this grade
- Weakest link
- Preparation — what was tested is not what is sold. Cochrane pooled the mineral trials it had — zinc and copper — and not magnesium L-threonate, so its null is about a different set of preparations.
- What was actually studied
- Cognitively healthy people aged 40 and over, 28 studies (n = 83000)
- Detail worth knowing
- B vitamins: little or no effect on global cognition to 5 years (SMD −0.03 to 0.06) and probably none at 5–10 years. Vitamin D3 400 IU/day with calcium in 4,143 participants: −0.1 MMSE points (95% CI −0.81 to 0.61) after a mean 7.8 years. Only two of the 28 studies had a mean baseline age under 60.
Everything above, and where it comes from
- Liu G, Weinger JG, Lu ZL, Xue F, Sadeghpour S (2016). Efficacy and Safety of MMFS-01, a Synapse Density Enhancer, for Treating Cognitive Impairment in Older Adults: A Randomized, Double-Blind, Placebo-Controlled Trial. Journal of Alzheimer's Disease 49(4):971–990
- Lopresti AL, Smith SJ (2025). The effects of magnesium L-threonate (Magtein) on cognitive performance and sleep quality in adults: a randomised, double-blind, placebo-controlled trial. Frontiers in Nutrition 12:1729164
- Rutjes AW, Denton DA, Di Nisio M, et al. (2018). Vitamin and mineral supplementation for maintaining cognitive function in cognitively healthy people in mid and late life. Cochrane Database of Systematic Reviews 2018(12):CD011906.pub2
The one thing worth doing
Leafy greens, nuts, seeds, legumes and wholegrains are where most people's magnesium comes from, and the amount in a normal week of those foods is larger than a capsule. Check the food before the form.
Regulatory note — this note has not yet been reviewed by a lawyer. In the United States, supplements are sold under DSHEA: a structure/function claim is not reviewed by the FDA before it appears, and the label must carry the words "This statement has not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease." In the EU and the UK, only health claims on the authorised register may be used in advertising, and a food or supplement advertisement may not say a product prevents, treats or cures a disease. Whether an evidence page that carried a purchase link would itself count as advertising under those rules is an open question that counsel, not this page, will answer; there are no purchase links here. Nothing on this page is medical advice.