Concentrations of magnesium and potassium in erythrocytes and plasma were determined in a population of 381 unselected elderly men and women, most of them in their eighties. The effects of biological factors (age, sex, weight) and a large set of pathological conditions, malignant or not, were examined. Analyses of variance showed a relation between age and concentrations of plasma potassium and between weight and concentrations of plasma magnesium. The chi-square test showed correlations between low concentrations of plasma magnesium and diabetes, abuse of alcohol and tobacco, and also between low values for erythrocyte magnesium and hypertension. Low values for plasma potassium were correlated with hypertension whereas high values were correlated with cardiovascular disease. Although some of the differences in the mean concentrations observed were statistically significant, these differences were always small. Most interesting was the distribution of the concentrations of the cations. This study shows that assays of both of these cations in erythrocytes were better than assays in plasma to evidence a deficiency. Indeed, about 20% of the studied population had low concentrations of both erythrocyte potassium and magnesium, whereas 2 and 10% had low values for plasma potassium and magnesium, respectively. This study underlines the large prevalence of magnesium and potassium deficiencies in the elderly, an observation we could not attribute to pathology or treatment. Routine electrolyte studies therefore appear to be justified in aged human subjects.