Reliable predictors of response for prostate cancer patients undergoing hormonal therapy are lacking. This study investigates the possibility that tumor proliferation rates might predict tumor behavior for these patients. The growth fraction of metastatic prostate cancer biopsy specimens obtained before androgen withdrawal therapy was evaluated by Ki-67 antibody immunohistochemical study to determine whether a higher tumor growth fraction was associated with a shorter time to tumor progression after therapy. The percentage of Ki-67-positive malignant epithelial nuclei in the primary tumors of 17 patients ranged from 1.7% to 7.5% (median, 2.5%). When patients were divided into two response groups according to the median time to progression, poor responders (time to progression less than 20 months) and good responders (greater than or equal to 20 months) had similar growth fractions (3.5 +/- 0.5% versus 3.1 +/- 0.6% Ki-67-positive nuclei, respectively). However, when patients were divided into two groups based on the median growth fraction, patients with a high growth fraction (greater than 2.5% Ki-67-positive nuclei) tended to have a shorter time to progression (median, 10 months) than patients with a low (less than 2.5%) growth fraction (median time to progression, 25 months), although statistical significance was not reached. Despite this interesting trend, Ki-67 immunostaining was not accurate to predict the time to progression in individual patients undergoing hormonal therapy. Reliable prediction of growth rates may require measurement of both cell proliferation and cell death rates.