A mineral that almost half of people in the US got too little of from food in 2005 to 2006. It lowers blood pressure a little, and blood sugar a little in people with or at risk of type 2 diabetes; in poor sleepers, trials improved a few sleep scores but not most, and for strength they found no clear gain.
Almost half of people in the US got less magnesium from food than they need in 2005 to 2006, and hard exercise raises what is lost in sweat and urine, which may lift the need by 10 to 20%. A supplement is measured by its magnesium, not the weight of its salt: in one trial, 893 mg of magnesium bisglycinate held 125 mg of magnesium.
My take
Core
Situational
Trying
Researching
Not taking
S-tier. Almost every cell runs on magnesium, and most diets fall short of it.
Dose
166 mga day
Timing
Eveningwith dinnerAlso 60 mg in the electrolytes, on waking and before training.
What to buyRead the label for the magnesium, not the weight of the salt. Mine says 166 mg magnesium (as magnesium glycinate); in one trial, 893 mg of magnesium bisglycinate held 125 mg of magnesium.
How to take itI take 166 mg in the evening after dinner, and my electrolytes add 60 mg a serving, so my day comes to 226 to 286 mg. The blood-pressure trials used a median of 365 mg a day.
What to expectNothing dramatic. Across adults, blood pressure fell by 2 to 3 mmHg, and in poor sleepers, 75 to 250 mg of magnesium a day improved a few sleep scores, such as insomnia severity or deep sleep, and left most others unchanged.
What the studies show
3 Strong1 Preliminary
Each segment is one studyBetterNo changeWorse
Strong: at least one pooled review or three randomised trials among the counted studies, and at least three in four of them better. Preliminary: everything else. Better is green in a Strong finding and ochre in a Preliminary one.
Better means a statistically significant change in the claimed direction at a normal dose. Each study counts once: a pooled review speaks for the trials inside it.
4of 5 studies betterof 5
Poor sleep and insomnia
155 people, Schuster 2025 · a small drop in insomnia scores at 250 mg of bisglycinate
4 of 5 studies better
In older adults with insomnia, magnesium cut the time to fall asleep by about 17 minutes in two small trials, at low certainty, and a review of adults found its trials contradictory. In three newer trials of adults with poor sleep, 75 to 250 mg of magnesium a day, as bisglycinate or L-threonate, improved a few sleep scores more than placebo, but most of their other sleep measures did not differ.
Strong
Sleep
What these 5 studies found. Pooled reviews: 1 better, 1 no change. Trials: 3 better.
In adults, its observational studies tied higher magnesium intake or blood levels to better sleep, but its randomised trials were contradictory, and it calls the effect of supplements uncertain.
In 155 healthy adults aged 18 to 65 with poor sleep, 250 mg of magnesium as bisglycinate a day, taken before bed for four weeks, lowered insomnia severity scores by 1.6 points more than placebo, a small effect; other sleep and mood scores did not differ.
In older adults with insomnia, the time to fall asleep was 17 minutes shorter than with placebo (2 of its 3 trials, 55 of its 151 people), and total sleep time 16 minutes longer, not significantly; it rates the evidence low to very low quality.
In 100 adults aged 18 to 45 with self-reported poor sleep, 2 g of magnesium L-threonate a day (145 mg of magnesium) for six weeks lowered sleep-related impairment scores more than placebo (p = 0.043, the paper testing one-tailed); sleep disturbance, restorative sleep and every Oura ring sleep measure did not differ.
Lopresti 2025 Frontiers in Nutrition
Randomised trial
100 adults aged 18 to 45 with self-reported poor sleep
2 g magnesium L-threonate (145 mg magnesium) a day · 6 weeks
In 80 adults aged 35 to 55 with self-reported poor sleep, 1 g of magnesium L-threonate a day (about 75 mg of magnesium), taken before bed for three weeks, improved deep and REM sleep scores on an Oura ring, and how people felt after waking, more than placebo; the change in insomnia scores from the start, rated sleep quality and total sleep time did not differ.
Hausenblas 2024 Sleep Medicine: X
Randomised trial
80 adults aged 35 to 55 with self-reported poor sleep
1 g magnesium L-threonate (about 75 mg magnesium) a day · 3 weeks
Found with a PubMed search: magnesium in the title with sleep or insomnia in the title or abstract, or magnesium in the title or abstract with sleep or insomnia in the title, reviews and trials, human studies. 77 results on 5 Oct 2026.
5of 5 studies betterof 5
Blood pressure
2,709 people, Argeros 2025 · the biggest drop in high blood pressure treated with medication
5 of 5 studies better
Magnesium lowers blood pressure a little: about 2 to 3 mmHg systolic and 1.5 to 2 diastolic across adults, at a median of about 365 mg a day, and most in people whose high blood pressure is treated with medication. In people with normal blood pressure, one review found a drop of about 3 mmHg systolic and another none, and the largest trial, in adults with high-normal blood pressure taking 360 mg a day, found none.
Strong
Heart
What these 6 studies found. The ring counts 5 of them: the other is a trial pooled in Behers 2023, counted through it. Pooled reviews: 5 better. Pooled in Behers 2023: 1 no change.
In mixed groups of adults, many with type 2 diabetes, metabolic syndrome or overweight, systolic pressure 2.5 mmHg and diastolic 1.6 lower than placebo or control (3,011 of its 5,226 people); systolic pressure fell in trials of people starting at 130 mmHg or more.
In adults with high or normal blood pressure, systolic pressure 2.8 mmHg and diastolic 2.1 lower than placebo, with no dose-response across 82.3 to 637 mg a day; 7.7 and 3.0 lower in people with high blood pressure on medication, and no significant drop in its 8 trials of people with normal blood pressure.
In adults with normal or high blood pressure, systolic pressure 2.0 mmHg and diastolic 1.8 lower than placebo, at a median 368 mg a day for a median three months; a dose of 300 mg a day or a duration of one month was sufficient to raise serum magnesium and lower blood pressure.
In adults with normal blood pressure, mostly healthy, 212 to 636 mg a day lowered systolic pressure by 2.8 mmHg and diastolic by 1.6 more than placebo, with high heterogeneity between its trials.
In its magnesium and placebo arms (461 of 698 healthy adults aged 30 to 54 with high-normal blood pressure), 360 mg of magnesium (as diglycine) a day left blood pressure no different from placebo at three and six months; the trial also tested calcium.
Yamamoto 1995 Annals of Epidemiology
Randomised trial
698 healthy adults aged 30 to 54 with high-normal blood pressure
In adults with high blood pressure, diastolic pressure 2.2 mmHg lower than control and systolic 1.3 lower, not significantly; it judges the evidence weak and probably due to bias, its 12 trials mostly of poor quality.
Dickinson 2006 Cochrane Database of Systematic Reviews
Meta-analysis
12 trials · 545 people
Found with a PubMed search: magnesium in the title with blood pressure or hypertension in the title or abstract, or magnesium in the title or abstract with blood pressure or hypertension in the title, reviews and trials, human studies, leaving out magnesium sulfate, pre-eclampsia, intravenous and pregnancy titles. 280 results on 5 Oct 2026.
0of 5 studies betterof 5
Strength and exercise performance
542 people, Wang 2017 · no gain in strength or power in athletes and active people
0 of 5 studies better
In athletes, active adults and untrained adults, magnesium did not clearly raise strength, power or endurance, and the one meta-analysis suggested a benefit only in older people and people with alcoholism. In the newest trial, 15 regular exercisers taking magnesium chloride for nine days had a slightly lower VO2max and sprint power than on placebo.
Preliminary
Performance
What these 5 studies found. Pooled reviews: 2 no change. Trials: 2 no change, 1 worse.
In adults without disease, athletes and non-athletes, it found the effects of magnesium on performance inconsistent and no evidence that it improves endurance; it finds little evidence for mineral supplements in sport, with magnesium a possible exception, its studies among the strongest in quality, and suggests weakly that higher doses may help strength and performance in younger people and moderate doses physical function in older people.
Heffernan 2019 Nutrients
Systematic review
663 adults without disease, athletes and non-athletes
In athletes and active adults, untrained adults, and older people and people with alcoholism, magnesium did not significantly raise peak torque, muscle strength or muscle power; benefits appeared more pronounced in the older and alcoholic groups, and not apparent in athletes and active people.
In 74 healthy middle-aged overweight women, 250 mg of magnesium (as oxide) a day for eight weeks left handgrip and knee-extension strength and a timed get-up-and-go test no different from placebo.
In 22 college-aged men and women, 350 mg of magnesium a day for 10 days left bench-press repetitions to failure no different from placebo (p = 0.06); soreness after eccentric bench pressing fell from baseline with magnesium and not with placebo.
Reno 2022 Journal of Strength and Conditioning Research
In 15 men and women who exercise regularly, in a crossover trial, 300 mg of magnesium chloride twice a day for nine days left VO2max (41.3 against 44.4 mL/kg/min) and mean 30-second sprint power (415 against 439 W) lower than placebo; a 10 km time trial did not differ.
Bomar 2025 Nutrients
Randomised trial
15 men and women who exercise regularly
300 mg magnesium chloride twice a day · 9 days
Found with a PubMed search: magnesium in the title with strength, exercise, sport, athletes or muscle in the title or abstract, or magnesium in the title or abstract with strength, performance, exercise, sport, athletes, muscle or physical in the title, reviews and trials, human studies, leaving out magnesium sulfate, pre-eclampsia, intravenous, pregnancy and cramp titles. 169 results on 5 Oct 2026.
3of 4 studies betterof 4
Blood sugar and insulin, with or at risk of type 2 diabetes
5,226 people, Mohammadi 2026 · small drops, in adults many with diabetes or at risk of it
3 of 4 studies better
In type 2 diabetes, magnesium lowered fasting blood sugar a little, by about 0.6 mmol/L in one review and 0.4 in the largest trial, where it was measured in a quarter of its 247 people, but HbA1c did not fall significantly in either. Across mixed groups of adults, many with or at risk of diabetes, it lowered fasting glucose, HbA1c and insulin resistance a little, though the newest review of people with diabetes or prediabetes found no significant drop in insulin resistance.
Strong
Metabolism
What these 4 studies found. Pooled reviews: 2 better, 1 no change. Trials: 1 better.
In mixed groups of adults, many with type 2 diabetes, prediabetes, metabolic syndrome or overweight, fasting glucose 3.6 mg/dL lower than placebo or control (2,514 of its 5,226 people), HbA1c 0.15 points lower (1,339 people) and insulin resistance (HOMA-IR) 0.48 lower (1,655 people); fasting insulin did not differ significantly.
In people with diabetes or prediabetes, neither fasting insulin nor insulin resistance (HOMA-IR) fell significantly against placebo; the effect depended on how high insulin and HOMA-IR started.
In adults with type 2 diabetes, at a median 360 mg a day for a median 12 weeks, fasting glucose 0.56 mmol/L lower than placebo; HbA1c 0.31 points lower, not significantly, and HDL cholesterol slightly higher.
In 247 adults with type 2 diabetes and HbA1c of 7% or more, 302 mg of magnesium (as 500 mg of oxide) a day for 12 months lowered fasting glucose by more than placebo, 0.4 against 0 mmol/L (67 of its 247 people); HbA1c, its main outcome, fell 0.25 points against none, not significantly (intention to treat).
Al-Maqbali 2026 Frontiers in Endocrinology
Randomised trial
247 adults with type 2 diabetes and HbA1c of 7% or more
302 mg magnesium (as 500 mg of oxide) a day · 12 months
Found with a PubMed search: magnesium in the title with glucose, insulin, HbA1c or diabetes in the title or abstract, or magnesium in the title or abstract with glucose, insulin, HbA1c or diabetes in the title, reviews and trials, human studies, leaving out magnesium sulfate, pre-eclampsia, intravenous, pregnancy and gestational titles. 315 results on 5 Oct 2026.
Dose
Thiago takes 166 mg in the evening, and his electrolytes add 60 mg a serving, so his day comes to 226 to 286 mg, under the upper limit of 350 mg a day for magnesium from supplements. Most trials gave 300 mg a day or more, at a median of 360 to 368 mg in three reviews, and one found 300 mg a day enough to raise blood magnesium and lower blood pressure. A supplement covers a common shortfall, which hard exercise may widen.
What people notice, what the upper limit means, and when to ask a doctor first.
CommonDiarrhoea and stomach upset
In trials for cramps, 11 to 37% of people on oral magnesium had side effects, mostly in the gut, such as diarrhoea, against 10 to 14% of controls. Unabsorbed magnesium draws water into the gut, magnesium oxide is used to treat constipation, and in one small trial glycinate was better tolerated than oxide.
The upper limit for magnesium from supplements is 350 mg a day; a 2026 review notes it has been questioned as conservative, the side effects mostly in the gut and rising with the dose. A review of 49 trials it cites, at up to 972 mg a day, found no serious side effects, only minor, passing ones.
Kidney disease. The kidneys clear extra magnesium, so in chronic kidney disease it can build up even at moderate doses; a 2023 review found no concern for severe excess when it was given to people with chronic kidney disease in clinical studies, under medical care.
Some antibiotics (tetracyclines and fluoroquinolones) or a bisphosphonate for bone loss. Magnesium, like calcium and iron, can bind them in the gut and lower how much is absorbed, so they are taken at different times, as each drug’s instructions say.
Magnesium bound to glycine, an amino acid, and Thiago’s form.
This page
Oxide
An inorganic salt. In a 2003 trial of 46 healthy adults, 300 mg of magnesium a day as oxide was absorbed less than citrate or an amino-acid chelate over 60 days, judged by the magnesium passed in urine, and no measure of magnesium in urine, blood, red cells or saliva differed from placebo on oxide. In a 1994 trial of 12 people missing part of the small intestine, it was absorbed as well as glycinate overall, but half as well in the four who absorbed it worst.
Magnesium bound to citric acid, an organic salt. In the same trial it raised blood magnesium most; a review finds organic salts slightly better absorbed than inorganic ones in most studies, which it calls hard to compare.
Magnesium bound to threonic acid, sold as Magtein. It carries little magnesium for its weight: in a 2025 trial, 2 g of it a day held 145 mg of magnesium.