Everstack Evidence

CC Small or few human trials. Promising, not proven. How we grade →Evidence grade · Last reviewed

Magnesium forms and dose: glycinate, citrate and oxide, and the 350 mg supplement limit

Bottom line: Magnesium has Grade C evidence for differences in absorption between forms, not sleep or calm benefits; the adult upper limit of 350 mg daily applies to supplements and medicines, and safety guidance is well established.

The safety guidance is solid: NIH sets the adult upper limit for magnesium from supplements and medicines at 350 mg a day, and the main drug interactions are well documented. The form comparisons are weaker. They come from small, short absorption studies that measure magnesium in urine or blood, not how people feel. A 2021 systematic review of 14 such studies found organic forms tend to be absorbed better than inorganic ones, with absorption depending on dose. No trial has compared forms head to head on sleep or calm. Grade C.

Published Sources checked 10 sourcesEverstack research teamHow we work

Reviewed . We sell some of what we grade.

Whole walnuts in their shells, a food source of magnesium
Whole walnuts in their shells, a food source of magnesium. Photo: Peppe Ragusa, CC0 1.0

Editorial section. Research summaries are not product claims and contain no product links or offers. A summary is never proof that any product works.

The question
Which magnesium form does the body absorb best, how much is sensible to take, what does the 350 mg upper limit cover, and who should check before taking magnesium?
Study design
Absorption studies and safety guidance. Walker 2003: randomized, double-blind, placebo-controlled trial in 46 healthy adults given 300 mg of elemental magnesium a day for 60 days as an amino-acid chelate, citrate or oxide. Lindberg 1990: lab solubility tests plus a single 25 mmol dose (about 600 mg) of citrate or oxide in healthy volunteers. Firoz and Graber 2001: urine magnesium in healthy volunteers given about 21 mEq (about 250 mg) a day of oxide, chloride, lactate or aspartate. Schuette 1994: double-blind randomized crossover trial in 12 patients who had part of the small intestine removed, comparing 100 mg of isotope-labeled magnesium diglycinate (glycinate) with oxide. Fine 1991: magnesium absorption in healthy volunteers at five doses taken with a meal. Blancquaert 2019: lab tests of 15 commercial products, then blood levels in 30 healthy adults after one dose of the best and the worst performer. Pardo 2021: systematic review of 14 bioavailability studies comparing forms. Costello 2016: expert perspective on the serum magnesium test. The NIH Office of Dietary Supplements fact sheet (intakes, upper limit, interactions), which draws on the 1997 Institute of Medicine Dietary Reference Intakes. [1][2][3][4][5][6][7][8][9][10]
What they found
Form: in Walker 2003, after 60 days citrate and the amino-acid chelate raised 24-hour urine magnesium more than oxide (P = 0.033), citrate gave the highest blood levels, and oxide did not differ from placebo. Lindberg 1990 found oxide almost insoluble in water and only 43% soluble even at peak stomach acid, while 55% of the citrate dissolved in plain water; urine magnesium rose far more after citrate. Firoz and Graber 2001 estimated oxide absorption at about 4%, with chloride, lactate and aspartate higher and similar to each other, so not every inorganic salt absorbs poorly. For glycinate itself the human data are thin: in Schuette 1994, absorption was about the same as oxide across all 12 patients (23.5% versus 22.8%), but twice as high in the four who absorbed oxide worst (23.5% versus 11.8%), and every patient tolerated the glycinate better. Product, not just form: in Blancquaert 2019 the 15 products varied widely, and two tablets both based mainly on oxide gave very different blood curves (area under the curve 6.87 versus 0.31 mM.min), so how a tablet is made matters too. Pardo 2021 concluded that organic forms appear more bioavailable than inorganic ones, that the share absorbed depends on dose, and that every form studied could keep normal levels in healthy people without a prior shortfall. Dose: in Fine 1991, the more magnesium taken at once, the smaller the share absorbed (from 65% at the lowest intake to 11% at the highest); magnesium from almonds was absorbed as well as a soluble salt, while an enteric-coated tablet was absorbed much less. Need and limit: adults need 310 to 420 mg a day from all sources (NIH ODS), and about half of Americans eat less than the average requirement (Costello 2016). The 350 mg upper limit counts supplements and medicines only, not food. Our numbers: 3 capsules of our magnesium give 275 mg as glycinate, and our multivitamin adds 50 mg as oxide, so taking both, as in Foundation, gives 325 mg a day, under the limit. [1][2][3][4][5][6][7][8][9][10]
Limitations
The absorption studies are small and short, mostly in healthy volunteers, and they measure magnesium in urine, blood or saliva. Those are markers of uptake, not proof of any effect on sleep, muscles or mood. Less than 1% of the body's magnesium is in blood serum (NIH ODS), and Costello 2016 argues that a normal serum test can miss a mild shortfall, so blood levels are an imperfect yardstick. Walker 2003 tested an amino-acid chelate that is not named as glycinate, and Schuette 1994 studied patients after bowel surgery, so neither shows that glycinate beats citrate in healthy people. Blancquaert 2019 was funded by Oystershell Laboratories, which also supplied the products, and the better of the two tablets tested in people (Ultractive Magnesium) is Oystershell's own; the authors state the company had no role in the design, analysis or decision to publish. Fine 1991 and Lindberg 1990 were NIH funded; Schuette 1994 was funded by the US FDA and NIH; Walker 2003's PubMed record tags it as supported by a source outside the US government, without naming the funder; funding for Firoz and Graber 2001 and Pardo 2021 was not stated in the abstracts. Splitting a dose across the day is a reasonable inference from Fine 1991, but no trial has tested it as an outcome.
What this does not mean
None of this shows that any magnesium form treats a disease, or that more is better. Our claim is structure/function only: supports muscle relaxation and a calm wind-down. Practical points from the sources: read the label for elemental magnesium, not the weight of the compound (in our capsules, 2,500 mg of magnesium glycinate gives 275 mg of magnesium). Count every magnesium supplement, antacid and laxative toward the 350 mg limit; food does not count. Diarrhea, nausea and cramps are the usual sign of too much, most often with carbonate, chloride, gluconate and oxide (NIH ODS). Serious toxicity is rare and has mostly been linked to laxatives or antacids giving more than 5,000 mg a day, with higher risk when the kidneys do not work well. Ask a clinician first if you have kidney disease, are pregnant or breastfeeding, or take bisphosphonates (keep them at least 2 hours apart from magnesium), tetracycline or quinolone antibiotics (take them at least 2 hours before or 4 to 6 hours after magnesium), diuretics, or a proton pump inhibitor for more than a year, which can lower magnesium. Not for children unless a clinician advises it; their limits are lower (65 to 350 mg depending on age).
Sources
  1. NIH Office of Dietary Supplements: Magnesium fact sheet for health professionals (intakes, 350 mg upper limit, interactions). ods.od.nih.gov/factsheets/Magnesium-HealthProfessional/
  2. Institute of Medicine. Dietary Reference Intakes for Calcium, Phosphorus, Magnesium, Vitamin D, and Fluoride. National Academies Press; 1997. doi.org/10.17226/5776
  3. Walker AF et al. Magnes Res. 2003;16(3):183-191. Citrate vs amino-acid chelate vs oxide, 46 adults, 60 days. pubmed.ncbi.nlm.nih.gov/14596323/
  4. Lindberg JS et al. J Am Coll Nutr. 1990;9(1):48-55. Citrate vs oxide, solubility and absorption. doi.org/10.1080/07315724.1990.10720349 Our summary: Magnesium bioavailability from magnesium citrate and magnesium oxide
  5. Firoz M, Graber M. Magnes Res. 2001;14(4):257-262. Four US commercial preparations. pubmed.ncbi.nlm.nih.gov/11794633/
  6. Schuette SA et al. JPEN J Parenter Enteral Nutr. 1994;18(5):430-435. Diglycinate vs oxide after ileal resection, 12 patients. doi.org/10.1177/0148607194018005430 Our summary: Bioavailability of magnesium diglycinate vs magnesium oxide in patients with ileal resection
  7. Fine KD et al. J Clin Invest. 1991;88(2):396-402. Absorption across five doses with a meal. doi.org/10.1172/jci115317 Our summary: Intestinal absorption of magnesium from food and supplements
  8. Blancquaert L et al. Nutrients. 2019;11(7):1663. 15 products in the lab, 2 tested in 30 adults. doi.org/10.3390/nu11071663 Our summary: Predicting and testing bioavailability of magnesium supplements
  9. Pardo MR et al. Nutrition. 2021;89:111294. Systematic review, 14 bioavailability studies. doi.org/10.1016/j.nut.2021.111294 Our summary: Bioavailability of magnesium food supplements: a systematic review
  10. Costello RB et al. Adv Nutr. 2016;7(6):977-993. Serum magnesium reference interval. doi.org/10.3945/an.116.012765 Our summary: Perspective: the case for an evidence-based reference interval for serum magnesium: the time has come

Common questions

Which form of magnesium absorbs better: glycinate, citrate or oxide?

In short absorption studies, citrate and amino-acid chelates raised urinary or blood magnesium more than oxide (Walker 2003; Lindberg 1990), and organic forms generally appear more bioavailable than inorganic ones (Pardo 2021). However, chloride, lactate and aspartate also absorbed better than oxide (Firoz 2001), tablet manufacturing changed blood curves even between two mostly-oxide tablets (Blancquaert 2019), and no trial has compared forms head to head on sleep or calm.

Does the 350 mg magnesium upper limit include food?

No. The NIH Office of Dietary Supplements 350 mg daily adult upper limit counts magnesium from supplements and medicines only, not food. In Fine 1991, the fraction absorbed fell from 65% at the lowest dose to 11% at the highest.

Related summaries

Spot an error? Email hello@everstackhealth.com and we will correct it within 48 hours.