In a randomized study 115 postmenopausal women with advanced breast cancer who were estrogen receptor-unknown or -positive were treated initially with tamoxifen or diethylstilbestrol (DES). Their pretreatment characteristics showed no significant difference. The frequency of response was identical with tamoxifen and DES, showing a complete response rate of 2% versus 2% and a partial response rate of 4% versus 8%, respectively; stable disease was present in 78% versus 73% of the patients, respectively. The median time to disease progression (5 vs 6 months) and median survival depending on initial hormone therapy (34 vs 35 months) were identical for tamoxifen and DES, respectively. Gastrointestinal toxicity was more frequent and more severe with DES than tamoxifen. Responses were seen with withdrawal of each agent and on crossover to the alternative agent. Our conclusions are that: DES and tamoxifen are equally effective in treating metastatic breast cancer in the postmenopausal patient who is estrogen receptor-positive or -unknown; withdrawal and crossover responses are seen with both agents; side effects are minimal but more frequent with DES; and on the basis of cost-effectiveness DES is the preferable agent.