Magnesium for depression, migraine, and dementia: 43 human studies reviewed - review

The Role of Magnesium in Depression, Migraine, Alzheimer's Disease, and Cognitive Health: A Comprehensive Review

Nutrients (MDPI), July 4, 2025, https://doi.org/10.3390/nu17132216

Péter Varga, Andrea Lehoczki, Mónika Fekete, … János Tamás Varga — Varga et al.

Summary by Claude Opus 5.5 - September, 2026

Takeaway: The evidence for magnesium is strongest for preventing migraines. Oral doses of about 400–600 mg per day for 2–3 months reduced migraine frequency and severity, especially migraine with aura. The evidence for depression is suggestive but mixed. For dementia, the data are observational and point to a "sweet spot" for blood levels rather than "more is better."

Study type: A peer-reviewed narrative review of 43 human studies (RCTs and cohorts, 2000–2025). There is no meta-analysis. The headline total of 219,224 participants is dominated by one UK Biobank cohort of 161,376 people.

Depression: About 8 of 12 trials reported improvement. Examples include magnesium oxide 500 mg/day for 8 weeks and magnesium chloride 248 mg/day for 6 weeks (PHQ-9 score fell 6 points, with effects starting in 2 weeks). Several trials found no benefit, including one in postpartum women and one in a large dietary cohort.

Vitamin D links: One trial in long COVID gave 1300 mg MgCl₂ plus 4000 IU vitamin D daily for 4 months and found that depression scores improved more than in controls. A trial in obese women gave 50,000 IU vitamin D weekly plus 250 mg Mg daily and found that inflammatory markers improved but depression scores did not. In US NHANES data, higher magnesium intake was linked to better cognition mainly in people with sufficient vitamin D. None of these designs can separate the effect of magnesium from the effect of vitamin D.

Migraine: Intravenous magnesium for acute attacks gave mixed results. It worked best in migraine with aura, where 2.7 patients needed treatment for one to benefit.

Dementia: Several cohorts show a U-shaped risk curve. Dementia risk was lowest at serum Mg of about 0.80–0.89 mmol/L, and both low and high levels were linked to roughly 30% more dementia.

What this does not show / limitations: The authors did not pool any data, so no overall effect size exists. Most trials were small and short and used different forms and doses of magnesium. Many did not report baseline Mg status, and serum Mg poorly reflects the magnesium inside cells. Depression outcomes were self-reported, so placebo effects are likely. The dementia findings are observational; for example, a Taiwanese cohort of magnesium oxide users could be confounded by why people took MgO, which is often used as a laxative. The text contains some errors: it describes Nazarinasab et al. as null, but its own Table 3 shows significant benefit. Achieved 25(OH)D levels are not reported for the vitamin D trials.

PDF

Related in VitaminDWiki