Abstract The stable isotope 26Mg was compared to the radioisotope 28Mg to estimate magnesium absorption in four healthy male subjects confined to a metabolic ward and consuming a constant diet for 146 days. Two isotopes tests were carried out in each subject on days 66 and 109 of the constant-diet period. In test 1, a solution containing 50 mg 26Mg and 30 μC 28Mg were taken orally during breakfast. In test 2, 50 mg 26Mg were administered orally followed three hours later by an i.v. injection of 20 μC 28Mg. True magnesium absorption was estimated by: (1) oral/i.v. ratios of 28Mg in plasma and urine, or oral/i.v. 26Mg/ 28Mg ratios in urine; (2) the difference between intake and fecal excretion of total dietary magnesium over a period of 10 days, or of a single dose of either isotope, corrected for endogenous fecal magnesium. Estimates made in individual subjects by different procedures based on data derived solely by use of 28Mg varied as much as estimates made by comparable procedures with data derived from either 26Mg or 28Mg. Measurements of 26Mg enrichment in urine and feces were made by neutron activation analysis which loses precision at enrichment levels below 10% above natural abundance. With the doses used in this investigation only fecal samples collected within 3–4 days and urine samples taken wihin 2–24 h contained adequately detectable 26Mg enrichment levels. Despite this limitation, 26Mg significantly expands the scope of investigation of magnesium absorption in man beyond that possible with the short lived 28Mg alone.