From June 1984 to March 1988, patients with newly diagnosed stage D2 prostate cancer were treated with protocol 1. This comprised oral hormonal agents either diethylstilbestrol diphosphate (Honvan: 300 mg/day) or estramustine phosphate (Estracyt: 560 mg/day), or chlormadinone acetate (Prostal: 100 mg/day), plus intravenous cyclophosphamide (CPM, 0.5-1 g/m2) every 3-4 weeks. From May 1988, protocol 2 was used in a randomized study of castration alone versus castration plus intravenous methotrexate (MTX, 20 mg/m2) every 2 weeks. Forty-nine of 53 patients who underwent the two protocols were evaluable for the response. The response rates according to the NPCP criteria were 92% (11/12) for Honvan, 100% (9/9) for Estracyt, 78% (7/9) for Prostal and castration plus MTX, and 80% (8/10) for castration alone. There were no significant differences among these treatments. The median response duration and survival time (months) were 16 and 44, respectively, for Honvan, 19 and 37 for Estracyt, 12 and 43 for Prostal, 11 and 15 for castration plus MTX, and 13 and 13 for castration alone. The short survival times of the castration alone and castration plus MTX groups were due to a short follow-up period. There were no statistical differences among the oral hormonal agent plus CPM groups. However, the 2-year survival rate (Kaplan-Meier method) was higher in the CPM and MTX groups than in the castration alone group. Survival was longer in the good performance status (P.S.) group than the poor P.S. group (p less than 0.05 by Wilcoxon test) and in the responders than the non-responders (p less than 0.01). Side effects were not excessive in the chemotherapy groups and patient compliance was good.