Magnesium L-threonate vs glycinate: what the human trials actually show

Magnesium L-threonate sells at a premium on one promise: this form reaches your brain. Behind it sit a rat study from 2010 and three human trials, all paid for by companies selling the ingredient. None measured magnesium in a human brain.

Glycinate is the boring option. Commodity chelate, decent tolerability, one good sleep trial with a small effect. Neither has ever been tested against the other, so the question is not which wins. It is which job you are doing.

What is threonate supposed to do that glycinate cannot?

Brain magnesium is held on a tight leash. Active transport across the blood-brain barrier keeps it steady, so pushing blood levels up does not move brain levels much. That is a genuine problem, and it is the real argument for a carrier molecule like threonate. Nobody has confirmed the carrier solves it in a person.

Magnesium runs over 600 enzymatic reactions, per de Baaij and colleagues, 2015, Physiological Reviews, which is why deficiency looks like everything and nothing. For the wider view, see the guide to magnesium forms.

The idea comes from Slutsky and colleagues, 2010, Neuron: aged rats given magnesium L-threonate showed better learning and memory and higher hippocampal synaptic density. Legitimate finding, legitimate reason to build a product, entirely rodent. No human study has measured brain or spinal fluid magnesium after threonate, and Liu and colleagues, 2016, said it cannot be done safely in people. So the mechanism that justifies the price is, in humans, unmeasured. Not disproven. Unmeasured.

What do the three human threonate trials actually show?

Three trials, 231 people enrolled, all funded by companies that sell the ingredient. Two were registered after they finished, so nobody outside can check what they set out to measure. The third and best was registered in advance and hit its primary endpoint at p=0.043. That is the whole human record.

Liu and colleagues, 2016, Journal of Alzheimer's Disease. 51 enrolled, 44 completed, 12 weeks, cognitive composite improved at p=0.003. Neurocentria funded it, the first author is Neurocentria's chief executive, and registration came about 15 months after the trial ended. The registered primary outcome was body magnesium status, the registered focus anxiety, mood and sleep. Cognition was a secondary outcome reported as the headline.

Hausenblas and colleagues, 2024, Sleep Medicine X. 80 randomized, 21 days, 1 g/day, only about 75 mg elemental. Deep sleep and REM scores improved. The objective minutes did not, and sleep duration, efficiency and wake time were null. A 2025 corrigendum disclosed late registration and that the manufacturer AIDP funded it, owns the patents, and pays three authors. The score moved. The clock did not.

Lopresti and Smith, 2026, Frontiers in Nutrition. 100 randomized, 6 weeks, 2 g/day, about 145 mg elemental. Prospectively registered, the strongest of the three, and the primary cognition composite was met at p=0.043. Caveats: more than 15 measures with no multiplicity correction, manufacturer-funded, manufacturer-supplied.

Industry funding does not make a result false. But registering a trial after it ends removes the one check that separates a planned finding from a chosen one. Promising, small, not yet independent.

How much magnesium are you actually getting?

Threonate is about 7.2 percent elemental magnesium as the commercial material is labeled, so a full 2 g serving delivers roughly 145 mg. The adult recommended daily allowance is 310 to 420 mg, and it counts food. Glycinate is about 14.1 percent, so it gives you twice as much per gram of material.

To land 200 mg of elemental magnesium you need about 2.8 g of threonate or about 1.4 g of glycinate. Twice the powder, at a patent premium. And that full 2 g serving, the dose the strongest trial used, sits below the daily requirement for every adult.

Fine, if repletion was never the goal. The problem is the person buying threonate as their magnesium supplement, paying the most per milligram and getting the least.

Is glycinate's own evidence any better?

Better registered, smaller effect. Glycinate's best trial is Schuster and colleagues, 2025, in Nature and Science of Sleep: 155 adults, prospectively registered, 250 mg elemental magnesium as bisglycinate. Insomnia Severity Index fell 3.9 points against 2.3 on placebo. That is p=0.049 and a Cohen's d of 0.2. Planned in advance, and small.

Say what p=0.049 means out loud. One participant landing the other way and this trial reports nothing. A d of 0.2 is the conventional floor for a detectable effect, the size most people would not notice in themselves. No other psychological outcome moved. The benefit did concentrate in people with lower baseline dietary magnesium, which is exploratory.

Now the unflattering part. Pajuelo and colleagues, 2024, Nutrients, gave 40 adults several forms and found bisglycinate produced no significant plasma magnesium rise at 1, 4 or 6 hours while oxide and citrate did, with comparable or greater gut complaints. That one came from a marketer of a competing form. Against it, Pardo and colleagues, 2021, Nutrition, reviewed 14 studies and found organic forms better absorbed than inorganic as a class.

Glycinate's only human head-to-head, Schuette and colleagues, 1994, JPEN, was null against oxide in 12 patients, 23.5 percent absorbed versus 22.8, though the chelate was better tolerated. Gentler, not obviously stronger. And if constipation is why magnesium came up, neither chelate is where the evidence points, a separate question covered in magnesium for constipation-predominant irritable bowel syndrome and in glycinate versus citrate.

What to do

1. Name the job first. Repletion and a cognition experiment are different projects with different products. If your fog arrives after meals and lifts later, read brain fog after eating before buying any magnesium.
2. For repletion, shop on elemental milligrams. Turn the bottle over and find the elemental number. Glycinate gets you there with half the material. Keep supplements at or under 350 mg a day.
3. If you try threonate, run it as an experiment, not as your magnesium. Match the strongest trial: 2 g a day for at least 6 weeks, one observable outcome, tracked in a symptom log that works. An effect this small is invisible without a baseline.
4. Clear the interactions. Separate magnesium from bisphosphonates by 2 hours, and take tetracycline or quinolone antibiotics at least 2 hours before or 4 to 6 hours after. Diuretics raise magnesium losses, and proton pump inhibitor (PPI) use beyond about a year carries a pooled relative risk of 1.43 for low magnesium, per Cheungpasitporn and colleagues, 2015, Renal Failure.

Common questions

Is magnesium L-threonate worth the extra money?

For repletion, no. You pay a premium for roughly half the elemental magnesium per gram. For cognition, it depends how much weight you give three manufacturer-funded trials, two of them registered after the fact.

Does threonate really get magnesium into the brain?

In rats, the data support it. In humans nobody knows, because nobody has measured it. Liu and colleagues, 2016, said the measurement cannot be done safely in people. Plausible biology, absent confirmation.

Can I take threonate and glycinate together?

Yes, and it is a reasonable way to separate the two jobs. Add the elemental milligrams from both, not the label weights, and keep the supplemental total at or under 350 mg a day.

Which one is better for sleep?

Glycinate has the better sleep trial, and "better" is doing light work. Schuster 2025 beat placebo by 1.6 insomnia points at p=0.049. The threonate trial moved sleep scores without moving measured minutes. Neither should make you confident.

Buy the evidence, not the mechanism, and never pay a premium for a measurement nobody has taken.