Magnesium: glycinate, citrate or oxide, and why the form matters

Citrate, on the only trial that ever compared them. Forty-six people took 300 mg a day of magnesium as citrate, as an amino-acid chelate, or as oxide, for 60 days.¹ The first two reached the bloodstream. Oxide came out no different from a dummy capsule.¹ Glycinate was not tested.

The trial that compared them

The comparison exists, it is small, and it is old.

Forty-six healthy people were assigned to one of three magnesium preparations, or to a dummy, all at 300 mg of elemental magnesium a day for 60 days.¹ Urine, blood and saliva were sampled at the start, 24 hours after the first dose, and at the end.¹

The two organic forms, citrate and the amino-acid chelate, were absorbed better than oxide by the end, judged by how much magnesium turned up in a day's urine.¹ Citrate produced the highest blood magnesium of any treatment, both after one dose and after two months.¹ Oxide produced no differences from the dummy capsule at all.¹

Four groups and 46 people works out to about a dozen per group.¹ That is thin. It is also the best direct evidence there is on the question the supplement aisle is organized around.

What "better absorbed" was measured with

The endpoints are worth separating, because they are not equivalent. Urine tells you how much magnesium got into the body and back out again. Blood tells you how much was circulating. Saliva is a convenience measure.

Red cell magnesium is the closest thing here to a tissue store, and it showed no differences between the groups.¹ So the form changed the numbers that turn over quickly and left the slow one alone, across 60 days.¹ Whether that matters depends on which number predicts an outcome, and this trial was not built to find out.

There is a further wrinkle in reading absorption off urine. Magnesium in urine is magnesium the kidney is throwing away. The same reading can mean good uptake, or a body with no room for more. Reporting it as absorption is the convention, and a convention is not a measurement of what reached the tissues.

The glycinate gap

Glycinate is the form most recommended online, and it does not appear anywhere in this evidence. Not in the comparison trial, not in the blood pressure reviews, not in the animal work.

The nearest proxy is the amino-acid chelate arm, since magnesium bisglycinate is an amino-acid chelate. That arm beat oxide on urinary excretion, and did not beat citrate on blood magnesium.¹

So it is fair to say glycinate belongs to a category that outperformed oxide on one marker. It is not fair to say a trial has shown it beats citrate. No trial here tried.

The confident three-way ranking that circulates online is not built on comparative trials. It is built on the chemistry of chelation, plus the absence of the complaints oxide produces. Neither of those is a measured outcome.

The widest comparison anyone ran was in rats

One study ranked many salts at once. It used 80 magnesium-depleted rats and a labeled tracer, feeding ten different salts including oxide, citrate, gluconate, lactate and aspartate.² Absorption ranged from 50% to 67%, organic salts came out slightly ahead of inorganic ones, and gluconate ranked highest.²

Excretion of the labeled magnesium ran from 0.20 mg to 0.33 mg, higher on the organic salts, and gluconate, lactate and aspartate held on to more than the inorganic salts did.²

Two things follow. In the study most likely to find a difference, the whole spread across every salt tested was 17 percentage points of absorption.² And these were rats, depleted for three weeks first, which is a design built to widen differences rather than to reproduce a person taking a capsule with dinner.

The ranking it produced does not match the shelf either. Gluconate came first, and almost nobody sells it.

Where absorption stops mattering

Blood pressure is the outcome most often measured with magnesium, and here the form question quietly disappears. The large analyses do not report which salt the trials used.

The most recent gathered 38 trials and 2709 participants, at doses from 82.3 mg to 637 mg, median 365 mg, over a median of 12 weeks.³ Systolic pressure fell by 2.81 mm Hg and diastolic by 2.05.³ In people with high blood pressure already on medication the systolic drop was 7.68, and in people low on magnesium it was 5.97.³ In people whose pressure was normal it did not reach significance.³

Then comes the finding that matters most here. There was no relationship between dose and effect at all, with every p value at or above 0.20.³ If tripling the dose does not change the outcome, a form that delivers somewhat more magnesium is unlikely to either.

The dose findings across reviews also contradict one another. A review of ten reviews and 8610 participants found the largest reductions at 400 mg a day or more, at 6.38 mm Hg systolic.⁴ A review restricted to people with normal blood pressure found the opposite: 360 mg a day or less gave 3.03 mm Hg, and courses beyond 3 months gave 4.31.⁵ An earlier analysis of 34 trials and 2028 participants landed at 2.00 mm Hg systolic and 1.78 diastolic, at a median 368 mg a day.⁶

That last one answers a question the others skip. 300 mg a day, or one month of it, was enough to raise blood magnesium, by 0.05 mmol/L over a dummy capsule.⁶ So the dose people actually take does arrive. The salt changes how much, not whether.

Systolic blood pressure reduction against placebo, across reviewsmm Hg

Argeros 2025, 38 trials-2.81 mm Hg

Alharran 2024, 10 reviews-1.25 mm Hg

Behers 2024, ≤360 mg/day-3.03 mm Hg

Zhang 2016, median 368 mg/day-2.00 mm Hg

Read together, the honest summary is a small blood pressure effect concentrated in people who are hypertensive or short of magnesium, on doses that do not line up from one review to the next.

What is still unknown

Whether glycinate does anything citrate does not. No trial in this set includes it.

Whether the laxative difference is real in trial data. It is the most common reason people switch forms, and none of these abstracts reports stool outcomes or side effects by salt.

Whether the absorption markers predict anything at all. The one trial that measured several found form effects on urine, blood and saliva, and none on red cell magnesium.¹

And which salts the outcome trials used. Every blood pressure analysis here sorts by dose and duration, never by form, which means the question this post is about has never been put to an outcome.³