116 patients aged 70 or over who were judged to have surgically resectable cancer of the breast were prospectively randomised to tamoxifen 20 mg daily or surgical resection. At a median follow-up of three years, local relapse or progression was seen in 15 (25%) of 60 patients in the tamoxifen group and 21 (37.5%) of 56 in the surgical arm. Distant metastases occurred in 8 (13%) in the tamoxifen group and in 10 (18%) in the surgical arm. There were 13 deaths in the tamoxifen group and 11 in the surgical arm, of which 8 and 9, respectively, were attributable to breast cancer. Disease-free survival did not differ between the groups.