Magnesium

Just under half of Americans take in less magnesium from food and drink than the median requirement for their age and sex — a measure of intake, not of deficiency, and people who take magnesium supplements on average clear the bar — and because the body defends blood levels by quietly draining bone and tissue stores, a standard blood test usually misses a shortfall. Its best evidence is for blood pressure — modest overall, larger in people who are deficient or already treated for hypertension — with a weaker sleep effect and promising but still-unproven longevity signals; the form you pick (glycinate, citrate, or threonate) mostly changes tolerability and cost, not how much magnesium you absorb.

Magnesium is a cofactor for over 300 enzymatic reactions, including the use of ATP (adenosine triphosphate, the cell's energy currency), DNA repair, and neuromuscular signalling. The dietary baseline has eroded over the last fifty years — modern soils are magnesium-depleted, refined grains lose most of the mineral during milling, and the "drink lots of water" message has accidentally pushed people toward reverse-osmosis-filtered water that strips minerals out. Replenishment via diet (leafy greens, nuts, legumes, whole grains) is the first move; supplementation closes the gap when diet can't.

The headline evidence

Blood pressure — Moderate (as a repletion effect). Pooling 34 trials (2,028 participants) at a median 368 mg/day, supplemental magnesium lowered blood pressure modestly overall — by about 2/2 mmHg.[1] A 2025 meta-analysis (38 trials, 2,709 participants) sharpens the picture: the overall effect is small (about −2.8/−2.1 mmHg), but it concentrates where physiology is already off — roughly −7.7 mmHg systolic in people treated for hypertension and −6.0 mmHg in those who are magnesium-deficient, with no significant effect in people whose blood pressure is normal, and no clear dose-response.[2] Two caveats sit inside that headline. The −7.7 mmHg comes from just six trials, and its 95% confidence interval runs from about −12.7 to −2.7 mmHg, so the true figure could be as small as about a third of the quoted one. And among people with hypertension it is mainly the treated who benefit: untreated hypertensives (nine trials) got no significant systolic reduction (−0.9 mmHg), though their diastolic pressure did fall slightly (−1.9 mmHg). For a population-scale intervention that's a real effect; for an individual it's roughly the size of cutting back on salt or losing a few kilograms — and it behaves as a repletion agent, largest in the deficient.

Genetic evidence is weaker than it first looks. A Mendelian-randomization study (60,801 coronary-disease cases) found that genetically higher serum magnesium tracked with about 12% lower coronary artery disease risk per 0.1 mmol/L increment (odds ratio 0.88, 95% CI 0.78–0.99).[3] (That 95% confidence interval is the range where the true effect most plausibly lies; because it stops just short of 1.0, the result is significant but only marginally so.) Three things temper it. The interval quoted is the fixed-effects one — the random-effects estimate (0.74–1.05) is not significant, although sensitivity analyses, including one that dropped a single outlying genetic variant, stayed significant. A later analysis using the same genetic instruments against a coronary-disease dataset about three times larger found no statistically significant association — its estimate pointed the same way, but its range included no effect — with signs that the genetic variants act through other pathways.[4] And the same Larsson paper found genetically higher magnesium associated with higher, not lower, systolic pressure, its authors concluding that they found no evidence that genetically higher magnesium lowers blood pressure — so this is not genetic backing for the effect above it.

A second under-appreciated route: drinking water itself. In a 16-year follow-up of 26,733 postmenopausal Swedish women, those living where the tap water was richer in both calcium and magnesium (magnesium averaging about 10 mg/L versus about 5 mg/L) had about 13% lower risk of ischemic stroke (hazard ratio 0.87). Separating the two minerals sharpens it: comparing the top third of water-magnesium exposure (10.6–13.8 mg/L) against the bottom third (median 4.5 mg/L) gives about 31% lower risk (hazard ratio 0.69), and only the water's magnesium, not its calcium, carried the association. Heart attacks showed no overall reduction (hazard ratio 0.93, a result compatible with no effect).[5] Reverse-osmosis filtration strips these minerals out (and isn't remineralised), working against this. See Water.

Other randomized and observational signals:

  • Sleep (Weak). Magnesium may bring modest benefits for selected subjective sleep outcomes in some adults, but the current best systematic review declines to go further: searching to mid-2026 and restricting to trials of magnesium as the only active ingredient, it found 12 randomized trials in adults and concluded that the evidence does not support oral magnesium as a routine treatment for insomnia, with results inconsistent and certainty low to very low.[6] The best recent randomized trial — 155 adults with poor sleep on 250 mg of elemental magnesium as bisglycinate, a dose that also delivers about 1.5 g of glycine — found a statistically significant but small improvement in insomnia severity (effect size ~0.2).[7] Magnesium is a physiological normaliser, not a sedative: any effect is small and appears largest where dietary intake is low. An earlier meta-analysis of three small trials in 151 older adults with insomnia found people fell asleep about 17 minutes sooner, at low to very low certainty on GRADE (the standard system for grading evidence quality),[8] so treat the sleep signal as possible but unproven. See Sleep and anxiety supplements.
  • Inflammation (Weak). The randomized evidence on C-reactive protein (CRP) is contradictory. One meta-analysis of 11 trials found no overall effect, with a reduction appearing only in people whose baseline CRP was already above 3 mg/L.[9] A second, smaller one found a significant overall reduction.[10] A 2025 update in metabolic-syndrome patients found a modest reduction.[11] The largest pooling to date — a 2026 review of 78 randomized trials, whose CRP analysis combined 17 comparisons in 777 participants — found no overall effect on CRP, though it did find a small drop in another inflammatory marker, interleukin-6, and it did not test whether people with high baseline CRP respond differently.[12] Effects on direct oxidative-stress markers are mixed, which suggests that any benefit is indirect — restoring normal physiology rather than scavenging free radicals.
  • Type 2 diabetes prevention (Strong for association; Moderate for treatment). This is one of magnesium's strongest signals. Across 40 cohorts (>1 million participants), each 100 mg/day of dietary magnesium tracked with 19% lower incident diabetes (RR 0.81, 95% CI 0.77–0.86).[13] A larger and more recent pooling — 35 diabetes cohorts with more than 1.2 million participants — puts a 100 mg/day increase in magnesium intake at about 6% lower risk (RR 0.94), so treat 19% as the high end of the range rather than the headline.[14] Randomized trials of treatment are more modest and population-dependent: insulin sensitivity (HOMA-IR) improves, but HbA1c (a marker of long-term blood sugar) and fasting-glucose effects emerge mainly in longer trials or in deficient patients, and a small 2024 randomised crossover cut against the repletion idea: it gave magnesium to insulin-treated patients selected for low serum magnesium, measured insulin sensitivity by the gold-standard clamp method, and found no benefit on its primary endpoint or on any glycaemic secondary — though with only 14 participants over six weeks, and a serum rise of just 0.05 mmol/L, the authors concede it may have been underpowered.[15] The honest read: strong for prevention/association, modest for treatment.
  • Stroke, heart failure, mortality (Moderate; observational). The same dose-response analysis links each 100 mg/day of dietary magnesium with lower stroke (about 7% lower, RR 0.93), heart failure (about 22% lower, RR 0.78), and all-cause mortality (about 10% lower, RR 0.90).[16] Two of those need discounting. The heart-failure estimate rests on 701 cases from three datasets — the thinnest cell in a million-person analysis — and a genetic study found no link between blood magnesium and heart failure, though its authors note it was too small to rule out a weak one.[17] And a later pooling built specifically for mortality, with about five times the deaths, puts the same increment at about 6% lower rather than 10%.[18] For a longevity audience this is the most relevant cluster — but it is observational dietary intake, which co-travels with potassium, fibre, and overall diet quality. The same paper hints at that confounding: where dietary magnesium tracked with lower mortality, magnesium supplements showed nothing — point estimates about 5% higher for all-cause and cardiovascular death and 11% higher for cancer death, none statistically significant. That rests on only two or three cohorts per outcome, too few to settle the question, but it gives no support to the supplement. Treat as a strong association with dietary patterns, not as evidence that a supplement prevents anything.
  • Migraines (Moderate). Magnesium has moderate evidence for migraine prophylaxis. The grade usually quoted for it — "probably effective", Level B from the American Academy of Neurology and the American Headache Society — comes from a 2012 guideline that the Academy formally retired in 2015, over safety concerns about a different treatment it recommended, and the Academy considers the conclusions of retired guidelines no longer valid.[19] European guidance grades magnesium Level C, listing it among third-choice preventive drugs with "only probable efficacy".[20] Neither guideline names magnesium oxide, and neither supports 400 mg/day: the retired American guideline recommends no magnesium dose, while the European one lists magnesium at 24 mmol — about 600 mg/day of elemental magnesium.
  • Cognition (Weak / preliminary). Magnesium threonate raises brain magnesium in animal models;[21] a 2025 randomized trial of 2 g/day in 100 healthy adults aged 18–45 who were dissatisfied with their sleep reported gains in working memory and faster reaction time.[22] The episodic-memory gain its abstract mentions did not differ significantly from placebo, and neither did five of the other six memory and thinking subtests; the two composite scores that make up the trial's headline score both failed, yet the headline score passed. It's a single trial of a branded form, and the headline "younger brain age" framing outpaces the evidence — but the working-memory signal is the most concrete human data the form has produced. One caveat to weigh: every human trial of this form for memory or sleep has been funded or supplied by its makers, and a 2024 sleep RCT (1 g/day) reported better ring-measured deep sleep and REM but no gain on the validated insomnia questionnaire it used. The "raises brain magnesium" claim derives from animal data, not human demonstration.

Why magnesium keeps appearing in longevity research

Weak / mechanistic. Most of the magnesium-and-ageing literature is mechanistic and preclinical, not outcome trials — worth understanding, but not the same evidentiary tier as the blood-pressure data above. The throughline is that magnesium sits upstream of several processes that go wrong with age, mapped comprehensively onto the hallmarks of aging in a 2024 review.[23] A few concrete examples:

  • Energy. ATP is biologically active only as a magnesium complex (Mg-ATP), so every reaction that spends cellular energy depends on magnesium status.
  • DNA integrity. Magnesium is a required cofactor for the main DNA-repair pathways and stabilises the double helix directly.
  • Cellular senescence. Human fibroblasts grown in magnesium-poor media accumulate senescence markers and exhaust their replicative lifespan faster — an in-vitro result, but a striking one.[24]
  • Anti-ageing signalling. A higher Magnesium Depletion Score — a clinical proxy for depletion, described under Testing below — tracks with lower circulating Klotho, a protein with cardiovascular- and kidney-protective effects that declines with age. Note the loop in that: up to two of the score's points come from kidney function, and Klotho is largely kidney-derived, so some of the association is built in.[25]

The one animal lifespan result worth flagging: adding soluble magnesium chloride to drinking water extended lifespan and reduced vascular calcification in a mouse model of accelerated ageing (progeria).[26] That rhymes with the human hard-water/stroke association, but it's a long way from proving magnesium extends human life. Treat this section as why magnesium is plausible as a longevity-relevant nutrient, not as evidence that supplementing it lengthens lifespan.

Form matters more for magnesium than for almost any other supplement

FormBest useAbsorptionNotes
Glycinate (bisglycinate)Sleep, anxiety, dailyNot establishedBetter tolerated than oxide in one small trial; no overall absorption advantage over oxide shown in people
CitrateDaily, mild constipationHighWell-absorbed; mildly laxative at higher doses
ThreonateCognition (preliminary)ModerateRaises brain magnesium in rats; brain uptake never measured in humans; expensive; low elemental yield
MalateDaytime, generalHighReasonable alternative; the fibromyalgia use was tested in a randomized crossover and was null in its blinded phase
TaurateCardiovascular nuanceHighTaurine-bound; the BP/arrhythmia use is mechanistic plausibility, no human outcome trials
ChlorideDaily, oralVariableUsually absorbed better than oxide, but results vary widely by product; the skin-spray use is the unsupported one
OxideConstipation onlyVery lowAlmost entirely laxative effect; poor systemic absorption
Sulfate (Epsom salt)Bath, occasional laxativeLow (oral)Skin absorption is minimal; the bath is mostly relaxation

The practical short version:

  • Glycinate in the evening if you want to try magnesium for sleep or anxiety — it was better tolerated than oxide in one small trial, though no overall absorption advantage has been shown.
  • Citrate if you want a daily generalist that's cheap and well-absorbed.
  • Threonate if you're specifically targeting cognition and accept that the evidence is preliminary and the cost is 5–10× higher. Its low elemental yield (a 2 g dose delivers only about 145 mg of elemental magnesium) means it's a poor choice for general repletion — pair it with a cheaper form if you also want to close the dietary gap.
  • Skip oxide for general supplementation — it's cheap but barely absorbed.

The organic-vs-inorganic gap is real but looser than it looks. Oxide is poorly absorbed and citrate clearly beats it,[27] but the figure usually quoted for oxide — about 4% — comes from a single urinary-excretion study at roughly 250 mg a day, in which chloride absorbed as well as the organic lactate and aspartate.[28] The share absorbed also falls steeply as intake rises: in one study that added a soluble salt to a standard meal in steps of up to about 970 mg of magnesium, it fell from about 65% at the lowest intake to about 11% at the highest, although the total amount absorbed kept rising.[29] And in a small stable-isotope crossover at 100 mg, in 12 patients who had part of the small intestine removed, oxide was absorbed at about 23% — no worse than glycinate.[30] A 2025 review still concludes that organic forms generally outperform inorganic ones, though it rates chloride's absorption as good alongside citrate's.[31] Head-to-head data among the organic forms remain thin, and no randomized trial has compared forms for sleep. The one modern direct comparison — a single 375 mg dose of oxide, citrate and bisglycinate in 40 adults, funded by the maker of a fourth, microencapsulated product tested alongside them — found plasma magnesium significantly above baseline at one time point after oxide and after citrate and at none after bisglycinate, but no significant difference between the forms in total exposure over 6 hours, so the absorption tiers above should be read as directional at best.[32]

Magnesium, vitamin D, and calcium

Magnesium doesn't act alone, and two interactions are worth getting right.

Vitamin D needs magnesium to work. The enzymes that convert vitamin D into its active hormone are magnesium-dependent, so low magnesium can blunt the response to vitamin D supplementation. A randomized trial found that magnesium supplementation lowered 25-hydroxyvitamin D in people whose levels were already high, while showing no consistent effect at low levels — the "raises the low, lowers the high" framing that circulates is only half supported.[33] Two limits are worth knowing: the participants were on average already vitamin-D-replete (arm means about 80 and 83 nmol/L, or 32 and 33 ng/mL) and almost all magnesium-sufficient, and magnesium's effect on the active hormone did not vary significantly with starting vitamin D level. If you're supplementing vitamin D and not getting the blood-level response you expect, magnesium status is one thing to check, but this trial is thinner support for that than it is usually made to carry. See Vitamin D.

Calcium and magnesium compete. They share absorption pathways, and modern diets heavy in dairy and calcium supplements have pushed the dietary calcium-to-magnesium ratio above the 1.7–2.6 window one research group has proposed as optimal.[34] A higher ratio went with a higher estimated 10-year cardiovascular risk score in one 2025 cross-sectional study of employees at an Iranian medical university.[35] The older evidence is weaker than it sounds: a US survey analysis of about 20,000 adults found only that exercise's cancer-mortality benefit appeared in the lower-ratio group, on an interaction test that was not significant,[36] and the one trial built around the ratio — magnesium dosed to bring it down to about 2.3 — measured vitamin D metabolites, not disease.[37] The 2:1 figure that circulates online is a 1989 expert recommendation, not a measured optimum, and no human study cited for the inflammation-and-bone claim supports it — the inflammation link traces to rat data, and a large ratio-and-fracture study in Chinese postmenopausal women found higher calcium intake protective only in women whose ratio was 1.7 or above.[38] Treat the whole idea as hypothesis-generating. The practical reading is unchanged: prioritise dietary calcium over high-dose calcium supplements (which carry their own cardiovascular signal — see Calcium), and don't let calcium crowd magnesium out.

If you're taking vitamin D for bone, the coherent stack is D3 + K2 + adequate magnesium: D raises calcium absorption, K2 directs that calcium into bone rather than arteries, and magnesium activates the D. The combination is mechanistically sound, but the bone evidence is thinner than that implies: pooled observational data link higher dietary magnesium to marginally higher bone density, with no fracture benefit at all — total fractures were flat, and the hip-fracture estimate actually pointed toward higher risk, though too imprecisely to mean much.[39] That analysis covers dietary intake only, with no supplementation trials in it. See Bone density.

Practical dosing

  • 200–400 mg/day of elemental magnesium covers the gap for most adults — though the top of that range sits above both upper limits for supplemental magnesium below, so it is not "more is better" territory.
  • The European Food Safety Authority (EFSA) upper limit for supplemental magnesium is 250 mg/day; the US upper limit is 350 mg/day. Food-based magnesium has no upper limit. (The reason: high-dose supplemental magnesium can cause osmotic diarrhea, which is the body's safety valve.)
  • Splitting the dose is commonly advised because the fraction absorbed falls as intake rises — in one study that added a soluble salt to a standard meal in steps of up to about 970 mg of magnesium, from about 65% at the lowest intake to about 11% at the highest.[40] Worth knowing that the total absorbed still rises at every step, so a single larger dose does deliver more magnesium than a single smaller one; splitting mainly helps tolerability.
  • Take glycinate or threonate in the evening if you're using it for sleep or relaxation. Citrate is anytime.

How to read the label

Supplement labels list the elemental magnesium content, not the total weight of the compound. A "magnesium glycinate 1000 mg" capsule delivers about 140 mg of elemental magnesium, because the salt is roughly 14% magnesium by mass. If a label claims 1000 mg of glycinate and 200 mg of elemental magnesium, the extra is almost certainly magnesium oxide in a "buffered" blend — the form this page tells you to skip. Check the supplement facts panel for the "Magnesium (elemental)" line.

Testing

Serum magnesium is insensitive for detecting body-store inadequacy — the body holds serum levels constant by pulling magnesium out of bone and intracellular stores. By the time serum magnesium drops, the deficit is severe. The international reference ranges are themselves contested: the MaGNet Global Magnesium Project — convened from an advocacy organisation, the Center for Magnesium Education and Research — argues the common lab cutoffs understate deficiency and proposes 0.85 mmol/L as the threshold for low magnesium.[41] Twelve nephrologists published a formal rebuttal, and on a 0.85 threshold about two-thirds of US adults would be classed deficient — which is the reason to treat the proposal as contested rather than settled.

  • There is no better routine test. Red-blood-cell magnesium is often promoted as the more sensitive alternative, but that superiority has not been demonstrated. The only systematic review of magnesium status biomarkers found serum, red-cell and urinary magnesium all responded to changes in intake, without showing any one to be better than the others,[42] and the most recent review concludes that no single biomarker reliably reflects whole-body magnesium status.[43] A 2026 clinical review treats total serum magnesium as the most practical measure and red-cell magnesium as an occasional adjunct when serum is normal but deficiency is suspected — noting that it is technically demanding and its reference ranges are not standardised between laboratories.[44] Read serum magnesium as a floor rather than a status reading, and do not chase the number upward: in 162,936 matched pairs of adults aged 50 and over, those with repeated readings below the conventional lower cutoff had about 33% higher dementia risk (hazard ratio 1.33) — but in a separate analysis of people without kidney disease, repeated readings above the normal range were also linked to higher risk (hazard ratio 1.25). These are health-records data from an older population, so illness that both shifts magnesium and raises dementia risk cannot be ruled out.[45]
  • Magnesium Depletion Score — a simple clinical score built from kidney function, diuretic and proton-pump-inhibitor use, and alcohol intake — was developed as an alternative to a blood draw. Its validation is thin: in the one study that tested it, in 77 adults already at high risk of deficiency, it identified deficiency only slightly better than serum magnesium did, and not clearly better than chance; the difference was never formally tested. In US population data a high score tracked with cardiovascular death, but its link with death from any cause was not statistically significant.[46] The strongest mortality association is in cancer survivors, where people with high scores were also about 23 years older and sicker.[47]
  • Symptoms of inadequacy are subtle: muscle cramps, low-grade fatigue, irritability, occasional palpitations. None are specific.

For most healthy adults with a normal diet who eat some nuts, leafy greens, and whole grains, magnesium status is adequate. The supplementation case is strongest in: heavy exercisers (sweat losses), older adults (declining absorption), people on diuretics (depletion), and people drinking very-low-mineral filtered water.

Cautions and interactions

  • Chronic kidney disease. The kidneys excrete excess magnesium, so as kidney function declines, magnesium can build up. The risk is narrower than it is usually stated: a nephrology review of magnesium given across all five stages of chronic kidney disease found no concerns for severe hypermagnesemia, and the cases that do occur are most often in older adults or people with kidney disease taking magnesium-containing laxatives or antacids.[48] If you have kidney disease, clear any magnesium supplement with the clinician managing it first.
  • Antibiotics (quinolones and tetracyclines). Magnesium chelates these drugs and reduces absorption. Separate doses by at least 2 hours.
  • Bisphosphonates (for osteoporosis): same problem. Separate by at least 2 hours.
  • Proton pump inhibitors (PPIs). Chronic use (usually more than a year) can lower magnesium absorption, and a supplement is a poor fix: in people who develop low blood magnesium on a PPI, even high-dose oral magnesium only partly restores blood levels, intravenous magnesium did not correct the deficit in the one study that tried it, and stopping the drug is the only reliable correction. If you are on a long-term PPI, the conversation to have with your prescriber is whether you still need it; if you do, a supplement can partly offset the loss but will not reliably correct it.[49]
  • Diuretics (especially loop and thiazide): increase magnesium loss; supplementation often warranted.

What's overrated

  • Transdermal magnesium — sprays, lotions, and Epsom-salt baths. The skin is an effective barrier to magnesium ions, and a 2017 review found no good evidence that topical application raises body magnesium meaningfully.[50] The Epsom-salt bath is relaxing, but it isn't a repletion route. Oral supplements remain the only well-supported way to raise magnesium.
  • Magnesium oxide for repletion — cheap and high in elemental magnesium by weight, but absorbed so poorly that most of it stays in the gut as a laxative. Fine if a laxative is what you want — a randomized trial found magnesium oxide 1.5 g/day improved chronic constipation in 71% of patients versus 25% on placebo[51] — but it's a poor choice for systemic repletion.
  • Magnesium for muscle cramps — one of the most marketed uses, and the evidence is essentially null. A Cochrane review concluded oral magnesium is unlikely to provide a clinically meaningful reduction in the idiopathic cramps of older adults; it found no randomized trials of magnesium for exercise-associated cramps at all, so it says nothing about cramps in younger, active people. For pregnancy cramps it reached no conclusion either way: it called the trials conflicting and said it was very uncertain whether pregnant women would benefit.[52] A later meta-analysis of four pregnancy trials found no benefit.[53]

Further reading

  • Magnesium supplementation and blood pressure — meta-analysis.[54]
  • Argeros Z et al. Magnesium Supplementation and Blood Pressure: A Systematic Review and Meta-Analysis of RCTs. Hypertension 2025.[55]
  • Fang X et al. Dietary magnesium intake and cardiovascular disease, type 2 diabetes, and all-cause mortality: a dose-response meta-analysis. BMC Medicine 2016.[56]
  • Larsson SC et al. Serum magnesium levels and risk of coronary artery disease: Mendelian randomisation study. BMC Medicine 2018.[57]
  • Helte E et al. Calcium and magnesium in drinking water and risk of myocardial infarction and stroke — a population-based cohort study. AJCN 2022.[58]
  • Garrison SR et al. Magnesium for skeletal muscle cramps. Cochrane Database Syst Rev 2020.[59]
  • Mori S et al. Magnesium oxide in patients with chronic constipation — randomized trial. J Neurogastroenterol Motil 2019.[60]
  • Rosanoff A et al. Suboptimal magnesium status in the United States: are the health consequences underestimated? Nutr Rev 2012.[61]
  • NIH ODS Magnesium Health Professional Fact Sheet.[62]
  • Dominguez LJ et al. Magnesium and the Hallmarks of Aging. Nutrients 2024.[63]
  • Killilea DW, Ames BN. Magnesium deficiency accelerates cellular senescence in cultured human fibroblasts. PNAS 2008.[64]
  • Dai Q et al. Magnesium status and supplementation influence vitamin D status and metabolism. AJCN 2018.[65]
  • Slutsky I et al. Enhancement of learning and memory by elevating brain magnesium. Neuron 2010.[66]
  • Lopresti AL, Smith SJ. Magnesium L-threonate (Magtein) on cognitive performance and sleep quality — a randomized, double-blind trial. Front Nutr 2025.[67]
  • Gröber U et al. Myth or Reality — Transdermal Magnesium? Nutrients 2017.[68]

— § —