Blood pressure (BP) and metabolic variables were determined initially and after 1, 2, 4, 6 and 10 years' treatment in two groups of hypertensive men (n = 53 each) randomized to bendroflumethiazide 2.5-5 mg/day or propranolol 160-320 mg daily. There was no significant differences in BP or metabolic variables between the two groups at entry. BP was reduced to the same degree by both treatments. Five men in the propranolol group and one man in the thiazide group developed clinically overt diabetes during follow-up. Fasting blood sugar increased slightly but significantly though equally in both groups. Oral glucose tolerance was initially impaired to the same degree in both groups but improved significantly during treatment with both drugs. Fasting insulin increased slightly but to the same degree. While serum potassium decreased significantly in the thiazide group, the total body potassium was unchanged in this group. In the propranolol group, serum potassium rose, while total body potassium decreased significantly. Serum urate increased in both groups, though slightly more during thiazide treatment. One case of gout was found in each group. There was no difference in serum lipids between the two groups. The finding in this long-term trial indicate that in middle-aged men with mild to moderate hypertension a low-dose thiazide diuretic like bendroflumethiazide is as effective and safe an antihypertensive agent as the beta-blocker propranolol is and that it does not induce diabetes. The total clinical picture favors the retention of thiazide diuretics as a first choice drug in hypertension.