A randomised phase II study was conducted with crossover design. Patients were randomly assigned to receive either three PC-SPES capsules orally three times a day or diethylstilbestrol (DES) 3 mg orally once a day. Prophylactic warfarin was administered. At clinical or prostate-specific antigen progression, patients received the other therapy. The study closed prematurely after PC-SPES was withdrawn from the market. Chemical analyses were performed on multiple lots of PC-SPES. Ninety patients were enrolled, of whom 85 were assessable for response. Prostate-specific antigen declines ≥ 50% were noted in 40% (95% CI, 25 to 56%) with PC-SPES and 24% (95% CI, 12 to 39%) with DES. Median response duration was not reached with PC-SPES, and was 3.8 months with DES. Median time to progression for randomly assigned patients was 5.5 months for PC-SPES and 2.9 months for DES. Common toxicities included mild fatigue, gynaecomastia and mastodynia. Five thromboembolic events occurred (one PC-SPES, four DES). Responses in the crossover phase were inconclusive. Four lots of PC-SPES had measurable quantities of DES, ranging from 0.01 to 3.1% of the dose used in the DES arm. Ethinyl estradiol was also detected in PC-SPES lots.