Magnesium is a physiological N-methyl-D-aspartate (NMDA) antagonist. The NMDA receptor may be involved in the pathogenesis of acute mountain sickness (AMS). In the present study, healthy subjects were randomized to receive either 400 mg of oral magnesium citrate (16 mmol) or matching placebo every 8 h for 5 days (prevention trial). Subjects then climbed to 4559 m in approx. 24 h and stayed there for 48 h. A Lake Louise Score <3 at any time was defined as the absence of AMS, whereas a score >6 (with ataxia, headache and nausea) was defined as a prevention failure. In a subsequent trial (treatment trial), subjects with a Lake Louise Score >6 (with ataxia, headache and/or nausea) were randomized to receive either 4 g of intravenous magnesium sulphate (16 mmol) or matching placebo. A decrease in the score >50% within 60 min was regarded as a treatment success. Dichotomous data were analysed using relative risk (RR) or odds ratio (OR), and continuous data using Student's t test or Wilcoxon's rank-sum test. In the prevention trial, data from 61 subjects (30 receiving magnesium and 31 placebo) were analysed. With oral magnesium, 20% of subjects had no AMS compared with 16.1% in the placebo group [RR (95% CI), 1.2 (0.4-3.6); where CI is confidence interval]. With magnesium, 40% were prevention failures compared with 35.5% in the placebo group [RR (95% CI), 1.13 (0.59-2.15)]. The mean time to failure and severity of AMS was similar between the two groups. With magnesium, 38.2% had loose stools compared with 11.8% in placebo group [RR (95% CI), 3.25 (1.18-8.97)]. In the treatment trial, 12 subjects received magnesium and 13 received the placebo. With intravenous magnesium, 25% were regarded as treatment successes compared with none in the placebo group [OR (95% CI), 9.71 (0.91-103.4)]. With magnesium, mean (+/- S.D.) scores decreased from 11.6 +/- 1.7 before treatment to 9.0 +/- 3.5 after treatment (P=0.009); scores remained unchanged in the placebo group. With magnesium, 75% of subjects experienced a transient flushing compared with 7.7% in the placebo group [RR (95% CI), 0.05 (0.01-0.25)]. In conclusion, oral magnesium does not prevent AMS. In subjects with established AMS, intravenous magnesium reduces the severity of symptoms to some extent, but this effect is of no clinical importance.