OBJECTIVE We confirmed the reported rate of prostate specific antigen (PSA) suppression after flutamide withdrawal in patients with metastatic prostatic carcinoma, increasing serum PSA and tumor progression following treatment with total androgen blockade (castration and flutamide). The value of clinical variables in predicting the rate of PSA decrease after flutamide withdrawal was assessed and adrenal androgen metabolism was correlated with the rate of PSA suppression following flutamide withdrawal. METHODS A total of 41 consecutive patients with metastatic prostatic adenocarcinoma and an increasing serum PSA while effectively castrated (plasma testosterone level less than 50 ng./ml.) who were receiving 250 mg. flutamide 3 times daily was evaluated prospectively before cessation of the flutamide. Responses were determined at 6 weeks. Only 2 of the 41 study patients (3%) had stable disease at 6 weeks, that is they had not met objective criteria for response or progression at analysis. RESULTS Of 39 patients studied 11 (28.2%, 95% confidence internal 14 to 45%) had a PSA decrease (more than 50% from baseline) following flutamide withdrawal and they were treated with initial complete androgen blockade. Median duration of PSA decrease was only 13 weeks (range 7 to 52), and 3 of the 11 patients had continued suppression of serum PSA concentrations at 12+, 13+ and 20+ weeks. The serum PSA decrease was associated with improved clinical symptoms, although objective regression of the disease was found in only 1 to 2 patients with measurable disease. No statistical correlation between endocrine studies or serum bombesin secretion and PSA decrease was found, although patients with a PSA decrease after flutamide withdrawal tended to have a lower dehydroepiandrosterone concentration than those with PSA progression. No correlation between known prognostic variables and decreased serum PSA after flutamide withdrawal was detected. CONCLUSIONS We confirmed the existence of the reported paradoxical PSA decrease in patients with androgen-independent carcinoma of the prostate, and that the delivery of simultaneous initial flutamide with castration predicts for PSA decrease. Individual patients appear to benefit from flutamide withdrawal although the overall impact was slight. The differences in frequency compared to those reported by others may be accounted for by patient selection and the number of patients receiving sequential castration therapy followed by flutamide.