A randomized clinical trial was performed to compare the efficacy of bilateral oophorectomy with that of tamoxifen at a dose of 10 mg twice daily in premenopausal women with metastatic breast cancer, and to examine the efficacy of each as a crossover treatment. Initial treatment responses were seen in ten of 27 patients (37%) treated with oophorectomy and seven of 26 patients (27%) treated with tamoxifen. The difference was not statistically significant. Crossover responses were seen in five of 15 patients (33%) treated with oophorectomy, including three responses in ten prior tamoxifen nonresponders; and two of 18 patients (11%) treated with tamoxifen. Time to progression distributions were not significantly different during initial treatment, and no significant differences in survival were noted. Thus, there was no overall disadvantage to the use of tamoxifen as opposed to oophorectomy as initial hormonal therapy, and a failure to respond to tamoxifen did not preclude a response to subsequent oophorectomy. Exploratory data analysis within subsets indicated consistent differential treatment effects in the visceral dominant patients. Of the 16 such patients treated with oophorectomy, eight (50%) experienced objective responses but there were no responses in the 14 patients treated with tamoxifen. In the nine visceral dominant crossover patients who had not responded to initial tamoxifen, three (33%) subsequently responded to oophorectomy. Time to progression distributions within the visceral dominant subset appeared to be better for the patients treated initially with oophorectomy. However, one must be very cautious in drawing conclusions from exploratory subset analyses, especially with the small sample size. Further studies would be required to test any hypothesis of differential organ site responsiveness.