Magnesium is involved in hundreds of enzyme reactions, nerve conduction, muscle contraction and the regulation of blood pressure and blood sugar. Intake from food is often below the reference amount in Western diets, which is the starting point for most of the marketing.

The supplement market sells it in many salts, glycinate, citrate, oxide, threonate and others, each with a story about absorption or a target organ. Almost none of those stories has a head-to-head trial behind it. The trials that exist mostly used whichever form was convenient and measured blood pressure, glucose, mood or sleep.

The claim most people buy it for, better sleep, has the weakest evidence on this page. The claims it is least sold for, blood pressure and glucose control, have the best.

Magnesium is a cofactor for the enzymes that make and use ATP, which puts it in every energy-dependent process in the body. It competes with calcium at cell membranes, which relaxes smooth muscle in blood vessels and dampens nerve excitability; that is the proposed route for both the blood pressure effect and the calming effect it is sold for.

In glucose metabolism it is needed for insulin signaling, and low magnesium is common in type 2 diabetes, where the kidneys lose more of it. Correcting a low level improving insulin sensitivity is a coherent mechanism, and the one with trial support.

For sleep, the proposed routes run through the GABA system and melatonin regulation. The trials testing that are few and small.

Blood pressure: a small, consistent effect. A 2016 meta-analysis of 34 double-blind placebo-controlled trials in 2,028 adults found oral magnesium lowered systolic blood pressure by about 2 mmHg and diastolic by just under 2 mmHg over a median of three months. That is a real effect and a small one, well short of what any blood pressure medicine does.

Glucose: supportive in people at risk. A 2021 meta-analysis of double-blind trials found magnesium reduced fasting glucose in people with diabetes and improved glucose after a glucose load and insulin-sensitivity markers in people at high risk of diabetes. Effects were on markers, not on whether anyone developed diabetes.

Depression: one open-label trial. A 2017 crossover trial in 126 adults with mild to moderate depression found six weeks of over-the-counter magnesium improved depression scores by six points on a standard scale. The trial was not blinded and had no placebo, which for a mood outcome is a serious limit; the authors' conclusion that magnesium is effective outruns the design.

Sleep: thin. A 2021 systematic review found three randomized trials, 151 older adults in total, all at moderate to high risk of bias. Pooled, magnesium shortened the time to fall asleep by about 17 minutes; total sleep time did not change significantly. The reviewers rated the evidence low to very low quality and said as much.

Oral magnesium's main effect at higher doses is diarrhea, which is also how the body limits how much it absorbs; magnesium salts are laxatives at pharmacy doses. In people with normal kidneys, excess is excreted. In kidney disease it accumulates, and high blood magnesium causes low blood pressure, slowed reflexes and, at the extreme, heart rhythm problems. That is the group for whom a supplement is a medical decision.

It also binds some antibiotics and other drugs in the gut and reduces their absorption, so timing relative to other medicines matters. Intravenous magnesium, used in hospitals for pre-eclampsia and some arrhythmias, is a different product and a different conversation.

Oral magnesium is sold as a dietary supplement in every major market, with no regulatory review of its claims. Nutrition authorities publish a reference intake and an upper limit for supplemental magnesium, which this site links to rather than copies.

Magnesium sulfate for injection is a licensed hospital medicine with specific indications. The existence of that medicine says nothing about the oral products on a shelf, and marketing that borrows its credibility is borrowing from a different compound and route.