Sleep
Dose: 250 mg/day elemental
Trial: N=155 RCT, 4 weeks, p=0.049
Insomnia Severity Index 1.6 points better than placebo, Cohen's d=0.20.
Read the trialGuide
Magnesium forms: match the form to the goal
There are 8 supplemental magnesium forms. Each is best for a different endpoint. Choosing the wrong form is why most people do not feel their magnesium.
8 supplemental formsSCF UL 250 mg/day elemental7.1x compound-to-elemental gap typical3 forms with real endpoint evidence
The label math
Why 2,200 mg of compound is only 310 mg of magnesium
Every bar below is the same 500 mg of compound. The only thing that changes is how much of that 500 mg is magnesium.
Swipe the chart sideways
Decision guide
Which magnesium form for what?
Six goals, six answers. Each card shows the dose the trial used and the trial that produced the number.
Bisglycinate
Cognition
Dose: 2 g/day compound, ~165 mg elemental
Trial: N=100 RCT, 6 weeks, p=0.043
NIH Total Cognition Composite favored threonate. Reaction time p=0.031.
Read the trialMigraine
Dose: 600 mg/day elemental
Trial: N=81 RCT, 12 weeks, p<0.05
Attack frequency fell 41.6% against 15.8% on placebo. Diarrhea in 18.6%.
Read the trialConstipation
Dose: 1,500 mg/day compound, 3 x 500 mg
Trial: N=34 RCT, 28 days, p=0.015
Overall improvement 70.6% against 25.0% on placebo. Colonic transit p<0.001.
Read the trialDepression
Dose: 248 mg/day elemental
Trial: N=126 RCT, 6 weeks, p<0.001
PHQ-9 net 6.0 points better adjusted, visible by week 2. Adjunct, not a substitute.
Read the trialInsulin resistance
Dose: 2,500 mg/day compound solution
Trial: N=63 RCT, 16 weeks, p=0.005
HOMA-IR 3.8 against 5.0. HbA1c 8.0% against 10.1%, p=0.04.
Read the trialDoses are stated as the trial stated them. Where a trial reported compound weight only, the elemental estimate is marked with a tilde. Four of these doses sit above the 250 mg/day supplemental upper level[14].
What actually gets absorbed
| Form | Elemental % | Absorption | Best evidence for | Watch out for |
|---|---|---|---|---|
| Oxide | 60.30% | Low, about 4% fractional | Constipation, N=34 RCT, p=0.015 | Laxative effect is the mechanism. No change in magnesium status at 300 mg/day |
| Chloride | 25.53% | High, same tier as citrate | Depression adjunct, N=126 RCT, p<0.001 | Diarrhea was the most common event. Salty taste in solution |
| Sulfate | 20.19% | Low by mouth | Eclampsia prevention, intravenous, N=10,141 | Extra osmotic load. Not a daily oral supplement |
| Citrate | 16.16% | High, best serum and saliva rise at 60 days | Migraine prevention, N=81 RCT, p<0.05 | 18.6% diarrhea at 600 mg/day, plus 4.7% gastric irritation |
| Malate | 15.54% | Not established, no randomized absorption study | Nothing survived the blinded phase | Energy claims with no human trial behind them |
| Glycinate | 14.10% | Medium, one crossover found no plasma rise | Sleep, N=155 RCT, p=0.049; pregnancy cramps, p=0.007 | Large pill load. The sleep effect is 1.6 ISI points |
| L-threonate | 8.25% | Not established, no head-to-head study located | Cognition, N=100 RCT, p=0.043 | Highest price per elemental mg. Fluid intelligence was null, p=0.953 |
| Taurate | n.a. | Not established | Blood pressure in 30 rats | No human trial located. PubChem publishes no molecular weight |
Absorption tiers are read from the cited comparison studies[6][7][15]. A systematic review of 14 absorption studies concluded only that inorganic salts absorb less well than organic ones[19].
The label trick: compound weight is not magnesium
2,200 mg magnesium glycinate = 310 mg magnesium
2,200 mg → 310 mg
Magnesium glycinate has a molecular weight of 172.42 g/mol and magnesium weighs 24.305 u, so 14.1% of the compound is magnesium[17]. The number on the front of the bottle is 7.1 times the number that matters. The Supplement Facts panel is required to declare elemental magnesium, not the weight of the whole compound[1].
| Form | Compound weight on a common label | Elemental % | Elemental mg delivered |
|---|---|---|---|
| Magnesium oxide | 400 mg | 60.30% | 241 mg |
| Magnesium chloride | 500 mg | 25.53% | 128 mg |
| Magnesium sulfate | 500 mg | 20.19% | 101 mg |
| Magnesium citrate | 1,000 mg | 16.16% | 162 mg |
| Magnesium malate | 1,000 mg | 15.54% | 155 mg |
| Magnesium glycinate | 2,200 mg | 14.10% | 310 mg |
| Magnesium L-threonate | 2,000 mg | 8.25% | 165 mg |
| Magnesium taurate | not stated | n.a. | n.a. |
Percentages are arithmetic from PubChem molecular weights and the magnesium atomic weight of 24.305 u. Taurate has no molecular weight published on its PubChem record, so its elemental share cannot be computed from a cited source.
Elemental density and absorption run in opposite directions. Oxide is 60.30% magnesium and absorbs at roughly 4%[7]. Citrate is 16.16% magnesium and produced the highest serum and salivary levels in a 46-person, 60-day randomized comparison, while oxide showed no difference from placebo[6]. Solubility explains it: citrate is about 55% soluble in plain water, oxide is close to insoluble[15].
Match the form to the goal
Eight forms, each with the strongest human evidence attached to it. Open the form you are considering.
Magnesium glycinate (bisglycinate)14.1% elemental
Best evidence: sleep. A 155-person randomized trial over 4 weeks moved the Insomnia Severity Index by 1.6 points more than placebo, p=0.049, Cohen's d=0.20, at 250 mg/day elemental[2] Second endpoint: pregnancy leg cramps, where 86.0% versus 60.5% of women reached a 50% cramp reduction, p=0.007, number needed to treat 3.9[3]
What it is for: sleep and cramps at a dose you can tolerate, plus general repletion if the pill count does not bother you.
What it is not for: a dramatic sleep fix. The effect is 1.6 ISI points. It is also not a high-density form: 14.1% means a 2,200 mg capsule stack carries 310 mg elemental[17] One 40-person crossover found no significant plasma rise after bisglycinate at all[1]
Magnesium L-threonate (Magtein)8.25% elemental
Best evidence: cognition. A 100-person, 6-week trial in adults aged 18 to 45 with poor sleep found the NIH Total Cognition Composite favored threonate, p=0.043, reaction time p=0.031[4] A separate 80-person, 21-day trial measured sleep outcomes[18]
What it is for: cognition. It is the only form with a positive cognition trial as its primary endpoint.
What it is not for: raising magnesium status. At 8.25% elemental, a 2 g compound dose carries about 165 mg elemental magnesium. Fluid intelligence was null in the same trial, p=0.953[4] It is also the most expensive magnesium per elemental milligram on the market.
Magnesium citrate16.16% elemental
Best evidence: migraine prevention. 600 mg/day for 12 weeks cut attack frequency 41.6% versus 15.8% on placebo, p<0.05[5] Citrate also produced the highest serum and salivary magnesium of the forms tested over 60 days[6] and ranks high on absorption in older comparisons[7]
What it is for: migraine prophylaxis and efficiently raising blood magnesium.
What it is not for: a calm stomach. The migraine trial reported 18.6% diarrhea and 4.7% gastric irritation at 600 mg/day[5]
Magnesium malate15.54% elemental
Best evidence: none that survives blinding. The only randomized study is a 24-person fibromyalgia crossover, and its blinded fixed-dose phase found no clear treatment effect. Pain reductions appeared only in the open-label dose-escalation phase[8]
What it is for: nothing proven. It is a reasonable general-repletion salt if you already own it.
What it is not for: energy or fibromyalgia, despite the marketing. There is no human trial supporting an energy claim for malate.
Magnesium tauratenot computable from a cited molecular weight
Best evidence: 30 rats. Magnesium taurate at 2 and 4 mg/kg/day for 4 weeks against cadmium-induced hypertension in Sprague Dawley rats[9] No human trial of magnesium taurate was located.
What it is for: nothing that human data support today.
What it is not for: blood pressure. Magnesium in general lowers systolic pressure by about 2 mm Hg across pooled trials, and the form used was usually not stated[1] That pooled effect is not evidence for taurate specifically.
Magnesium oxide60.30% elemental, the highest of the eight
Best evidence: constipation. 1,500 mg/day compound for 28 days produced 70.6% overall improvement versus 25.0% on placebo, p=0.015, with colonic transit time p<0.001[10]
What it is for: constipation, and nothing else.
What it is not for: correcting magnesium status. Fractional absorption is about 4%[7] and 300 mg/day for 60 days produced no difference from placebo on magnesium markers[6] Oxide is close to insoluble in water[15] The highest elemental percentage on the shelf is attached to the worst absorption.
Magnesium chloride25.53% elemental
Best evidence: depression. 248 mg/day elemental for 6 weeks improved PHQ-9 by a net 6.0 points adjusted, 95% CI 7.9 to 4.2, p<0.001, with effects visible within 2 weeks[11] Secondary endpoint: insulin resistance, HOMA-IR 3.8 versus 5.0, p=0.005, over 16 weeks[12]
What it is for: cheap repletion, low mood as an adjunct, and insulin sensitivity signals. Absorption is in the same tier as citrate[7]
What it is not for: a substitute for depression treatment. It was tested alongside usual care, not against it.
Magnesium sulfate20.19% elemental, anhydrous basis
Best evidence: intravenous, not oral. The 10,141-woman Magpie trial cut eclampsia from 1.9% to 0.8%, a 58% relative reduction[13] That is a hospital drug protocol.
What it is for: clinical obstetric use, and bath salts.
What it is not for: daily oral supplementation. No oral longevity indication exists, and the European upper-level opinion singles sulfate out for the extra osmotic load[14]
What to avoid
Taurate. No human trial exists. The whole blood pressure story rests on 30 rats[9]. Price it accordingly.
Malate sold for energy. The blinded phase of the only randomized study reported no clear treatment effect[8]. The energy claim comes from the malic acid half of the name, not from a trial.
Oxide sold for anything except constipation. It has the constipation trial[10] and it has 4% fractional absorption[7]. Using it to correct magnesium status means swallowing the laxative effect for almost none of the mineral.
Mixed-form sleep formulas. Blends that list four salts and a total compound weight usually deliver a sub-therapeutic elemental dose. The sleep trial used 250 mg/day elemental from 1,786 mg of compound[2]. If a sleep blend cannot show you an elemental figure in that range, it is not running the studied protocol.
Split what you do take. Relative uptake is higher across several small doses than one large one[16].
Dose ceiling and drug interactions
- 250 mg elemental per day is the supplemental upper level set by the EC Scientific Committee on Food, uncertainty factor 1.0[14].
- 360 to 365 mg/day is the lowest-observed-adverse-effect level. The adverse effect is diarrhea[14].
- Separate by 2 hours from fluoroquinolone and tetracycline antibiotics and from oral bisphosphonates. Magnesium binds all three and cuts their absorption[1].
- Kidney impairment: cut the dose by half or more, or skip supplemental magnesium entirely. Reduced clearance raises the risk of hypermagnesaemia[1].
Safety and interactions
The EC Scientific Committee on Food set the tolerable upper intake level for supplemental magnesium at 250 mg/day elemental, with an uncertainty factor of 1.0. The no-observed-adverse-effect level was 250 mg/day and mild diarrhea appeared at roughly 360 to 365 mg/day. Toxic hypermagnesaemia was reported only above 2,500 mg/day oral[14]. Several of the trials above run above the ceiling on purpose: 300 mg for cramps, 600 mg for migraine, 1,500 mg of oxide compound for constipation.
Kidney impairment. Reduced kidney function clears magnesium poorly and raises the risk of hypermagnesaemia. Anyone with chronic kidney disease should not self-dose magnesium[1].
Drugs that deplete magnesium. Loop and thiazide diuretics increase renal magnesium loss. Proton pump inhibitors taken for a year or more have been associated with hypomagnesaemia[1].
Drugs magnesium binds. Magnesium forms complexes with fluoroquinolone and tetracycline antibiotics and with oral bisphosphonates, cutting their absorption. Separate the doses by at least 2 hours[1].
Diarrhea is the dose-limiting side effect for every form. At 600 mg/day of citrate it hit 18.6% of participants[5].
How our top-rated magnesium products stack up
Ranked by our six-pillar overall score. Form and endpoint fit are read from the product label data we hold. No brand pays for placement.
Magnesium Bisglycinate
Thorne HealthTech
Nitralis®
DoNotAge
Magnesium Glycinate Powder
BulkSupplements
Pure Vitamin D3, K2 & Magnesium
DoNotAge
Basic Nutrients 2/Day
Thorne HealthTech
Two-Per-Day Multivitamin (capsules/tablets)
Life Extension
See the full ranking: Best Magnesium Supplements 2026.
Frequently asked questions
Which magnesium is best for sleep?
Bisglycinate has the only randomized sleep trial with a positive primary endpoint: 155 adults, 4 weeks, Insomnia Severity Index 1.6 points better than placebo, p=0.049, Cohen's d=0.20, at 250 mg/day elemental. The effect is real and small.
Is magnesium threonate worth the price?
It is the only form with a positive cognition trial as its primary endpoint: 100 adults, 6 weeks, NIH Total Cognition Composite p=0.043. It is also the weakest form by elemental content at 8.25%, so it is a poor choice for correcting magnesium status. Buy it for cognition or not at all.
What is the difference between compound and elemental magnesium?
Compound weight is the whole salt. Elemental is the magnesium atom inside it. A label reading 2,200 mg magnesium glycinate carries about 310 mg elemental magnesium, a 7.1-fold gap. The Supplement Facts panel declares elemental. Front-of-bottle copy often does not.
Can I take too much magnesium?
Yes. The EC Scientific Committee on Food set a tolerable upper intake level of 250 mg/day of supplemental elemental magnesium. Mild diarrhea appeared around 360 to 365 mg/day. Toxic hypermagnesaemia was reported only above 2,500 mg/day oral. Kidney impairment lowers the safe ceiling.
Does magnesium taurate work?
No human trial of magnesium taurate was located. The blood pressure evidence is 30 rats given 2 to 4 mg/kg/day for 4 weeks. PubChem does not publish a molecular weight for the compound, so the elemental percentage cannot even be computed from a cited source.
Related pages
Corrections and source queries: [email protected].
References
- [1] NIH Office of Dietary Supplements. Magnesium, Health Professional Fact Sheet. https://ods.od.nih.gov/factsheets/Magnesium-HealthProfessional/
- [2] Schuster et al. 2025. Magnesium bisglycinate and insomnia severity, N=155, 4 weeks. Nature and Science of Sleep 17:2027-2040. https://pmc.ncbi.nlm.nih.gov/articles/PMC12412596/
- [3] Supakatisant and Phupong 2012. Magnesium for pregnancy leg cramps, N=86. Maternal and Child Nutrition. https://pmc.ncbi.nlm.nih.gov/articles/PMC6860204/
- [4] Lopresti and Smith 2026. Magnesium L-threonate and cognition, N=100, 6 weeks. PMID 41601871. https://pubmed.ncbi.nlm.nih.gov/41601871/
- [5] Peikert et al. 1996. Magnesium citrate for migraine prophylaxis, N=81. Cephalalgia. PMID 8792038. https://pubmed.ncbi.nlm.nih.gov/8792038/
- [6] Walker et al. 2003. Bioavailability of magnesium citrate, amino-acid chelate and oxide, N=46, 60 days. PMID 14596323. https://pubmed.ncbi.nlm.nih.gov/14596323/
- [7] Firoz and Graber 2001. Bioavailability of commercial magnesium preparations. PMID 11794633. https://pubmed.ncbi.nlm.nih.gov/11794633/
- [8] Russell et al. 1995. Magnesium and malic acid in fibromyalgia, N=24 crossover. PMID 8587088. https://pubmed.ncbi.nlm.nih.gov/8587088/
- [9] Magnesium taurate and cadmium-induced hypertension in 30 rats. PMC6435948. https://pmc.ncbi.nlm.nih.gov/articles/PMC6435948/
- [10] Mori et al. 2019. Magnesium oxide for chronic constipation, N=34, 28 days. PMID 31587548. https://pubmed.ncbi.nlm.nih.gov/31587548/
- [11] Tarleton et al. 2017. Magnesium chloride for depression, N=126. PLoS ONE. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0180067
- [12] Rodriguez-Moran and Guerrero-Romero 2003. Magnesium chloride and insulin sensitivity, N=63. Diabetes Care. PMID 12663588. https://pubmed.ncbi.nlm.nih.gov/12663588/
- [13] Magpie Trial Collaborative Group 2002. Magnesium sulphate for pre-eclampsia, N=10,141. Lancet. PMID 12057549. https://pubmed.ncbi.nlm.nih.gov/12057549/
- [14] EC Scientific Committee on Food 2001. Tolerable Upper Intake Level of Magnesium, SCF/CS/NUT/UPPLEV/54 Final. https://ec.europa.eu/food/fs/sc/scf/out105_en.pdf
- [15] Lindberg et al. 1990. Magnesium bioavailability from magnesium citrate and magnesium oxide. PMID 2407766. https://pubmed.ncbi.nlm.nih.gov/2407766/
- [16] Schuchardt and Hahn 2017. Intestinal absorption and factors influencing bioavailability of magnesium. PMID 29123461. https://pubmed.ncbi.nlm.nih.gov/29123461/
- [17] PubChem compound record, magnesium glycinate CID 84645. https://pubchem.ncbi.nlm.nih.gov/compound/Magnesium-Glycinate
- [18] Hausenblas et al. 2024. Magnesium L-threonate and sleep, N=80, 21 days. PMID 39252819. https://pubmed.ncbi.nlm.nih.gov/39252819/
- [19] Pardo et al. 2021. Bioavailability of magnesium food supplements, systematic review of 14 studies. PMID 34111673. https://pubmed.ncbi.nlm.nih.gov/34111673/