Some clinicians contend that hypomagnesemia is a common problem in patients receiving diuretic therapy and that routine serum magnesium determinations may be indicated in such patients. We determined serum magnesium (Mg++) levels in 354 patients with uncomplicated hypertension. No significant difference was observed in the mean Mg++ between the 245 diuretic-treated patients and the 109 patients not receiving diuretics, 0.965 vs 0.97 mmol/L (1.93 vs 1.94 mEq/L). When analyzed by type of diuretic, there were statistically significant differences in the mean serum Mg++ concentrations between those receiving thiazides, 0.94 mmol/L (1.87 mEq/L); those receiving no diuretics, 0.97 mmol/L (1.94 mEq/L); and those receiving triamterene-containing diuretics, 1.01 mmol/L (2.01 mEq/L). These absolute differences, however, were clinically quite small, and hypomagnesemia was uncommon. Neither patient age, the duration of diuretic use, nor the serum potassium level correlated with Mg++. With respect to dose, those receiving 100 mg/d of hydrochlorothiazide had the lowest Mg++ concentrations and the greatest prevalence of hypomagnesemia (12%), defined as Mg++ less than 0.75 mmol/L (1.5 mEq/L). Serum Mg++ need not routinely be determined in patients with uncomplicated hypertension who are receiving triamterene-containing diuretics or low-dose (50 mg/d or less) hydrochlorothiazide.