This study was conducted to determine the similarity of response rates and safety produced by weekly paclitaxel with or without oral estramustine in patients with metastatic hormone-refractory prostate cancer. Between December 1998 and December 1999, 163 patients were randomized to receive 28-day cycles of paclitaxel 100 mg/m2 on days 2, 9, and 16 plus estramustine 280 mg orally 3 times a day on days 1-3, 8-10, and 15-17, or to receive paclitaxel 100 mg/m2 alone on days 1, 8, and 15. Objective response was defined as a > oe = 50% decrease in prostate-specific antigen (PSA) maintained for 4 weeks with stable or improved performance status. Response rates included 37 partial responses for paclitaxel/estramustine (47%) and 22 partial responses for paclitaxel (27%; P < 0.01). Median duration of response was 15.1 months for paclitaxel/estramustine and 15.5 months for paclitaxel; median survival was 16.1 months and 13.1 months, respectively (P = 0.049). Common toxicities for both treatments included neutropenia, gastrointestinal events, neuropathy, and asthenia. Thromboembolic events were more frequent in the paclitaxel/estramustine arm (no prophylactic anticoagulants). The rate of PSA decline for paclitaxel/estramustine was almost 2 times that of paclitaxel (47% vs. 27%), with acceptable toxicity. Multivariate analysis of prognostic factors affecting survival was not significant for treatment arm (P = 0.08). Although the incidence of thromboembolic events appeared to be increased in the paclitaxel/ estramustine arm, the addition of estramustine was responsible for a 20% increase in the rate of PSA decline. Neither treatment arm had significant impact on quality of life as measured by the Functional Assessment of Cancer Therapy-Prostate quality of life questionnaire. This study produced encouraging data; further studies of paclitaxel/ estramustine are recommended.