Most patients with metastatic carcinoma of the prostate have osteoblastic bone metastases and nonmeasurable pelvic disease. These features cause patients to be at high risk for myelosuppression after cytotoxic chemotherapy and make it difficult to evaluate response to treatment. A critical review of larger trials that have sought to assess the role of chemotherapy in treatment of carcinoma of the prostate leads to the following conclusions: (1) Although the aim of treatment is palliation, most trials have tried to evaluate tumor response rather than the more appropriate endpoints of quality and quantity of survival for all treated patients. (2) Criteria that have been used for tumor response are variable and contain large inherent errors; most patients who are labeled as "responders" are described as being "objectively stable," but this category may be a manifestation of slowly progressive disease rather than a response to treatment. (3) There is no evidence that chemotherapy causes a meaningful prolongation of survival. (4) Chemotherapy adds considerable toxicity, and reported trials have not adequately assessed its overall impact on quality of life. Because of these factors there is little evidence that chemotherapy provides palliation for patients with prostatic carcinoma, and it should not be regarded as part of standard management. Selected patients who are symptomatic and no longer responding to hormones may be considered for trials of chemotherapy. Future trials should randomize patients to chemotherapy or supportive care, with assessment of quality and quantity of survival for all randomized patients by an observer who is unaware of the treatment.