A summary of published research, not medical advice. See the Disclaimer.
Magnesium supplements are sold as different compounds, such as oxide, citrate, bisglycinate and L-threonate, and this page sets the human studies on those forms side by side.
How we grade evidence · Last reviewed: November 2026
What you need to know
Few studies compared magnesium forms directly, and most of those measured blood or urine magnesium, which are lab markers and do not show a health benefit. Citrate raised urine magnesium more than oxide in four comparisons, with weak evidence, but blood results disagreed. In the one long head-to-head trial, 164 people used citrate, oxide or sulfate at 450 mg a day for 24 weeks, and no form changed blood pressure or arterial stiffness compared with placebo (weak evidence). Sleep, constipation and other effects were tested one form at a time against placebo, so any comparison between forms on those effects is indirect. Case reports describe dangerously high blood magnesium after oxide laxatives in older people with kidney dysfunction, and one death after magnesium chloride. In 15 healthy adults, aspartate lowered levothyroxine exposure by 12%, and citrate lowered it by 7%, which was not significant. Trials of specific forms that stated elemental magnesium used 250 to 504 mg a day.
| Comparison | Verdict | What was measured | Studied in |
|---|---|---|---|
| Organic forms absorbed better than inorganic forms | May help (weak evidence) Review of 14 studies, but one study found chloride equal to organic salts |
Lab marker | |
| Citrate raises urine magnesium more than oxide | May help (weak evidence) Four comparisons agree, one small study found no gap, no health benefit shown |
Lab marker | Healthy adults, and overweight or slightly obese adults |
| Citrate raises blood magnesium more than oxide | Unclear: trials disagree (weak evidence) Two studies favour citrate, one found no gap, one favoured oxide |
Lab marker | Healthy adults, and overweight or slightly obese adults |
| Bisglycinate or amino-acid chelate absorbed better than oxide | Unclear: trials disagree (weak evidence) Better on urine in one trial, no overall gap in another, no plasma rise in a third |
Lab marker | Healthy adults, and adults with ileal resection |
| Chloride, lactate and aspartate absorbed better than oxide | Unknown (very weak evidence) One comparative study, size and randomisation not stated in the abstract |
Lab marker | Normal volunteers |
| Citrate, oxide or sulfate changes blood pressure or arterial stiffness | May make little or no difference (weak evidence) One trial of 164 people, no form beat placebo at 450 mg a day |
Measured | Overweight and slightly obese adults, mean age 63 |
Organic against inorganic forms. May help (weak evidence): a systematic review of 14 studies concluded that inorganic forms appeared less bioavailable than organic ones and that the percentage absorbed depended on dose [1]. The same review concluded that all forms could maintain normal levels in healthy people without a prior deficit, which could not be assured in older people or people with illnesses or low levels [1]. One study found magnesium chloride, an inorganic salt, equal to lactate and aspartate [2]. The grade reflects that conflict and an abstract that gives neither study sizes nor a pooled result.
Urine magnesium, citrate against oxide. May help (weak evidence): citrate raised urine magnesium more than oxide after a single load in normal volunteers [3], after 60 days at 300 mg/day in 46 healthy adults [4], and over 24 weeks at 450 mg/day in 164 adults [5]. After a single 400 mg dose in 14 healthy men, urine magnesium rose significantly with citrate and not with oxide [6]. In 11 people given 486 mg/day for 2 weeks, urine magnesium rose by about the same amount with either salt: 77 to 79 mg/day on an empty stomach and 92 to 96 mg/day with meals [7]. The four studies with stated sizes enrolled 235 people, and two of the five tested a single dose. The claim is about the marker itself, so the lab-marker rule does not apply, but a higher urine level does not show a health benefit.
Blood magnesium, citrate against oxide. Unclear: trials disagree (weak evidence): citrate gave the highest serum level in 46 healthy adults, and oxide did not differ from placebo [4]. Citrate also gave higher plasma levels at 4 and 8 hours in 14 healthy men [6]. In the 24-week trial, the rise in plasma magnesium was similar with citrate, oxide and sulfate [5]. In 61 young women over 10 days, oxide raised ionised magnesium while citrate raised the ionised share and lowered total serum magnesium, and those authors judged oxide the most bioavailable [8].
Bisglycinate and amino-acid chelate against oxide. Unclear: trials disagree (weak evidence): an amino-acid chelate and citrate both beat oxide on 24-hour urine magnesium at 60 days in 46 healthy adults, though citrate led on serum [4]. That abstract does not say whether the chelate was bisglycinate. In 12 people with ileal resection, absorption of a labelled 100 mg dose did not differ overall (23.5% for diglycinate, 22.8% for oxide), although diglycinate gave 23.5% against 11.8% in the 4 people who absorbed oxide worst [9]. In 40 healthy adults on a low-magnesium diet, plasma magnesium did not rise significantly within 6 hours after bisglycinate, while it rose at 1 hour after oxide and at 4 hours after citrate; each product was tested against its own baseline in this part of the analysis [10].
Chloride, lactate and aspartate against oxide. Unknown (very weak evidence): one study in normal volunteers found fractional absorption of 4% for oxide and higher, equal absorption for chloride, lactate and aspartate [2]. The abstract gives neither the number of volunteers nor any randomisation.
Blood pressure and arterial stiffness. May make little or no difference (weak evidence): 164 overweight or slightly obese adults received citrate, oxide, sulfate or placebo at 450 mg/day for 24 weeks, and no form changed pulse wave velocity or blood pressure compared with placebo [5]. This is the only head-to-head trial of forms on a health measure that we found. It is one trial, and it was not a test of whether one form beats another on these outcomes.
These rows are not head-to-head. Each form was tested in its own trials, so any comparison between rows is an indirect comparison across separate trials and cannot show that one form is better. The Magnesium article grades magnesium in general.
| Effect and form | Verdict | What was measured | Studied in |
|---|---|---|---|
| Insomnia severity, bisglycinate | May help a little (weak evidence) One trial of 155 adults, 1.6 points more than placebo, small effect |
Self-reported | Adults aged 18 to 65 reporting poor sleep |
| Sleep, L-threonate | May help (weak evidence) Two trials of 80 and 100 adults, the sleep-ring result was not repeated |
Measured and self-reported | Adults aged 18 to 55 with self-reported sleep problems |
| Constipation, oxide | May help (weak evidence) Two small placebo-controlled trials, almost all women |
Self-reported | Adults with chronic constipation, almost all women |
| Blood pressure, lactate | May help (weak evidence) Two small trials, one lost significance by 24 weeks |
Measured | Adults with implanted defibrillators, women with metabolic syndrome |
| Birth and maternal complications, citrate | Probably makes little or no difference (moderate evidence) One trial of 829 women, rates similar to placebo, wide ranges |
Hard outcome | Pregnant women at higher risk |
| Pain and depressive symptoms in fibromyalgia, magnesium with malic acid | May make little or no difference (weak evidence) Only one randomised trial among 11 primary studies |
Self-reported | People with fibromyalgia |
| Absorption, malate and acetyl taurate | Unknown (very weak evidence) Rat and mouse studies only, no human absorption study found |
Lab marker | Rats and mice |
Bisglycinate and insomnia severity. May help a little (weak evidence): in 155 adults reporting poor sleep, 250 mg/day of elemental magnesium as bisglycinate lowered Insomnia Severity Index scores by 3.9 points at week 4, against 2.3 points with placebo (p = 0.049, Cohen's d = 0.2) [11]. Other psychological outcomes did not differ. It is one small trial with a self-rated outcome and no objective sleep measure.
L-threonate and sleep. May help (weak evidence): in 80 adults aged 35 to 55, 1 g/day for 21 days improved deep and REM sleep scores on a sleep-tracking ring compared with placebo [12]. In 100 adults aged 18 to 45, 2 g/day for 6 weeks improved sleep-related impairment, with no group differences in sleep disturbance, restorative sleep or sleep outcomes on the same type of ring [13]. Both papers later had corrections published, and the correction notices carry no abstract, so we could not read what changed.
Oxide and constipation. May help (weak evidence): in 33 women analysed, 0.5 g of oxide three times a day for 28 days gave overall improvement in 70.6% against 25.0% with placebo, without a significant change in complete bowel movement response [14]. In a trial of 90 patients, 93% of them women, 1.5 g/day gave overall improvement in 68.3% against 11.7% with placebo [15]. These are laxative results. They say nothing about whether oxide is a better or worse source of magnesium.
Lactate and blood pressure. May help (weak evidence): in 50 people with implanted defibrillators, 504 mg/day of elemental magnesium lowered systolic pressure at 12 weeks (117.7 against 126.3 mmHg), and the gap lost significance at 24 weeks in the 45 who continued [16]. Diastolic pressure did not change. A trial in 58 women with metabolic syndrome found a significant drop in systolic but not diastolic pressure [17].
Citrate in pregnancy. Probably makes little or no difference (moderate evidence): in 829 pregnant women at higher risk, 300 mg of citrate from 12 to 20 weeks until delivery did not lower the combined perinatal outcome (18.4% against 18.0% on placebo, adjusted odds ratio 1.10, 95% CI 0.72 to 1.68) or the combined maternal outcome (12.0% against 9.7%, adjusted odds ratio 1.29, 95% CI 0.83 to 2.00) [18]. It is a single trial with wide ranges, which holds the grade at moderate.
Magnesium with malic acid. May make little or no difference (weak evidence): a summary of seven systematic reviews found 11 primary studies, only one of them randomised, and concluded that the combination made little or no difference to pain and depressive symptoms in fibromyalgia [19].
Malate and acetyl taurate. Unknown (very weak evidence): after a single dose in rats, the area under the curve was highest with malate and second highest with acetyl taurate, and oxide and citrate had the lowest bioavailability [20]. In mice, acetyl taurate raised brain magnesium at all doses tested [21]. The rat result for citrate runs against the human studies above, which is one reason animal rankings do not carry over to people. We found no human absorption study of malate or taurate.
| Side effect | Verdict | How often vs placebo | Source | Studied in |
|---|---|---|---|---|
| Poorer tolerance with oxide than with diglycinate | May happen (weak evidence) One crossover trial of 12 patients, no rates in the abstract |
Not reported | [9] | Adults with ileal resection |
| Diarrhoea and other minor gut effects, forms not separated | May happen (weak evidence) Reviews pooled forms, 7 of 10 studies found no difference in diarrhoea |
11% vs 10% up to 37% vs 14% in cramp trials; 7 of 10 studies found no difference in diarrhoea | [22][23] | Mostly older adults and pregnant women with cramps; adults |
| Dangerously high blood magnesium with oxide or chloride | Unknown (very weak evidence) Case reports and one hospital cohort, no comparison with other forms |
Case reports only; 23% of 320 hospital patients prescribed oxide, no comparison group | [24][25][26] | Older adults, hospital patients |
We found no study that compared gut side effects between common forms in a way that could rank them. Oxide is widely prescribed as a laxative [24][26], and these reports have no comparison with other forms, so they cannot show that oxide is riskier than other forms at the same dose.
This lists what studies and labels report. A pharmacist or prescriber can say whether it applies to a specific person.
| Combined with | What happened | Verdict | Source | Studied in |
|---|---|---|---|---|
| Levothyroxine, aspartate and citrate | Aspartate lowered thyroxine exposure (AUC) by 12% in 15 healthy adults; citrate lowered it by 7%, not significant | May happen (weak evidence) One open-label crossover study, no direct test between the two forms in the abstract |
[27] | 15 healthy adults |
| Gabapentin, oxide | 1 g of oxide lowered gabapentin exposure (AUC) by 43% and bioavailability by 32% in 13 healthy adults | May happen (weak evidence) One open-label crossover study, other forms not tested |
[28] | 13 healthy adults |
This lists what studies and labels report. A pharmacist or prescriber can say whether it applies to a specific person.
These are the amounts used in the studies. They are not recommendations.
Doses below are elemental magnesium where the abstract says so. Absorption studies used 300 mg/day for 60 days [4], 400 mg/day for 5 days before a single 400 mg test dose [6], 486 mg/day for 2 weeks [7] and a labelled 100 mg dose [9]. The head-to-head trial used 450 mg/day for 24 weeks [5]. The bisglycinate sleep trial used 250 mg/day of elemental magnesium, assessed at week 4 [11]. The lactate trials used 504 mg/day [16] and 84 mg twice a day [17]. The pregnancy trial used 300 mg of citrate daily [18].
The L-threonate trials gave 1 g/day and 2 g/day of the compound [12][13], and the constipation trials gave 1.5 g/day of oxide [14][15]. Those figures are the weight of the compound, so they cannot be set directly beside elemental doses.
Across 48 trials (n=2,131) of oral magnesium in adults without severe disease, blood and urine magnesium rose with dose and time, levelling off at about 300 and 400 mg/day, or after about 20 and 40 weeks [29]. That analysis pooled the response to oral magnesium and did not report results by form.