This study examines the outcome following relapse for 176 patients who had been entered into a randomised trial comparing adjuvant cyclophosphamide, methotrexate and 5-fluorouracil (CMF) with no adjuvant therapy (controls). Relapse has occurred in 65/144 (45%) of the CMF group and 111/158 (70%) of controls (P < 0.0001). 123/176 patients received endocrine treatment after relapse with higher response rates (38 vs. 18%, P < 0.05) and longer time to progression (23 vs. 19 weeks, P = 0.03) for controls. 94/176 received chemotherapy after relapse again with higher response rates (47 vs. 23%, P = 0.05) and longer time to progression (17 vs. 9 weeks, P = 0.03) for controls. Despite this, survival after relapse was the same for the two groups (median 16 months). However, on subgroup analysis, postmenopausal patients who had received adjuvant CMF had shorter survival (P = 0.03). These results suggest that prior adjuvant therapy should be a stratification factor in clinical trials in advanced disease.