Increasing awareness of the hazards of diuretic therapy has cast doubt on the appropriateness of its uniform use as initial treatment for hypertension. To establish the relative efficacy and acceptability of alternative forms of initial therapy, prazosin and hydrochlorothiazide were prospectively compared. In a one-year study, 62 patients with a diastolic blood pressure greater than or equal to 100 mm Hg were randomly allocated to receive one of the two agents. Twenty of 32 (63 percent) patients receiving prazosin and 17 of 30 (57 percent) receiving hydrochlorothiazide satisfied the year-long study requirements. For the most part, the dropout rate was not drug related. The percentages of subjects completing the study with the initial drug, without addition of a second drug, were identical in both groups (prazosin: 10 of 32 [31 percent] and hydrochlorothiazide: nine of 30 [30 percent]). In both treatment groups, blood pressure declined similarly and significantly (p less than 0.001) by the end of the study. The blood pressure of subjects receiving prazosin decreased from 150.3/105.8 mm Hg to 135.3/90.3 mm Hg, whereas the blood pressure of subjects receiving hydrochlorothiazide declined from 147.3/103.8 to 130.0/89.1 mm Hg. A reduction in diastolic blood pressure to less than 95 mm Hg was achieved in 80.0 percent of patients started with prazosin and in 78.6 percent of those started with hydrochlorothiazide. A further reduction to less than 90 mm Hg was achieved in 45.0 and 64.3 percent, respectively (not significant). Patients who completed the study with single-drug therapy in both treatment groups showed nonsignificant decreases in mean potassium levels from initial to final readings when compared with baseline readings (prazosin: 0.21 meq; hydrochlorothiazide: 0.41 meq). However, patients receiving prazosin had 11.5 percent of their readings less than or equal to 4.0 meq, whereas those receiving hydrochlorothiazide had 38.1 percent of their readings similarly distributed (p less than 0.05). Furthermore, four patients receiving hydrochlorothiazide had potassium levels less than or equal to 3.5 meq; none of those receiving prazosin had levels in that range. At baseline, no patient in the study had potassium levels of less than 4.0 meq. Prazosin-treated patients whose initial fasting blood glucose was less than 110 mg/dl had 3.7 percent (one of 27) of subsequent measurements greater than or equal to 110 mg/dl. These abnormal levels were found 18.2 percent (six of 33) (not significant) of the time in similar hydrochlorothiazide-treated subjects. Symptomatic side effects were primarily mild and transient in those receiving either drug alone.(ABSTRACT TRUNCATED AT 400 WORDS)