The magnesium loading test is a useful tool in the diagnosis of magnesium deficiency. In order to establish a reference range in normal subjects, 88 healthy men and women aged between 18 and 66 years were given 30 mmol magnesium intravenously during eight hours as a loading test, urine was collected from start of infusion for 24 h for measurement of magnesium excretion. The magnesium mean retention was 6.3 +/- 10.3% of the loading dose, and the 0.025 and 0.975 fractiles were -19.5% and 27.5%, respectively. There was no significant difference between the sexes or in the different age groups studied. There was no correlation between the magnesium retention and serum magnesium or with basal urinary magnesium excretion. An excess excretion of magnesium was observed the postload day compared to baseline, but the excretion 24 and 48 h after the magnesium loading were closely correlated, suggesting that 24 h urinary sampling is sufficient. In order to examine the reproducibility of the test 23 of the subjects underwent two magnesium loading tests 4 weeks apart. The mean difference between two repeat magnesium loading tests was 2.0% with a SD of 8.1% and a 95% confidence interval of -1.6-5.5%. Normal saline did not affect baseline magnesium excretion. Concomitant administration of ethanol or physical exercise caused greater variation in magnesium excretion, whereas furosemide was without effect. The 8 h magnesium loading test with 24 h urine sampling seems to be fairly reproducible, is adequate for clinical use, but the normal range is wide.