We compared the efficacy and safety of the gonadotropin-releasing hormone analogue, leuprolide (1 mg subcutaneously daily), with diethylstilbestrol (DES, 3 mg by mouth daily) in patients with prostate cancer and distant metastases (Stage D2) who had not previously received systemic treatment. Initial therapy (leuprolide or DES) was continued for as long as an objective response was noted; cross-over to the alternative arm occurred at the time of disease progression or intolerable adverse reactions. Ninety-eight patients were randomly assigned to leuprolide, and 101 to DES. Suppression of testosterone and dihydrotestosterone and decreases in acid phosphatase were comparable in the two groups. Patients receiving DES experienced more frequent painful gynecomastia (P less than 0.00001), nausea and vomiting (P = 0.02), edema (P = 0.008), and thromboembolism (P = 0.065) than those receiving leuprolide. The leuprolide group reported more "hot flashes" (P = 0.00001). Overall, 86 per cent of the leuprolide group had an objective response (complete response, 1 per cent; partial response, 37 per cent; stable disease, 48 per cent), as compared with 85 per cent of the DES group (complete, 2 per cent; partial, 44 per cent; stable, 39 per cent). Actual survival rates at one year were 87 per cent for the leuprolide group and 78 per cent for the DES group (P = 0.17). We conclude that leuprolide offers an important alternative treatment that is therapeutically equivalent to and causes fewer side effects than DES for the initial systemic management of metastatic prostate cancer.