A retrospective analysis of prognostic factors in 214 consecutive node-positive (N+) operable breast cancer patients, receiving Melphalan + 5-fluorouracil adjuvant chemotherapy between 1980 and 1984 was performed. Median follow-up was 95 months. Actuarial disease-free interval (DFI) and survival (S) were determined according to age, menopausal status, histology, size of primary tumor (T), multifocality, tumor location, hormonal receptor status, number of N+, size of N+, tumor spread in axillary fat, and interval between surgery and onset of adjuvant chemotherapy. On univariate analysis two factors were prognostic for DFI and S: number of N+ and T size. A comparison between traditionally classified T1 and T2 patients revealed no significant difference, but when the cut-off point was shifted from 2 cm to 3 cm, T size represented a highly significant prognostic factor. In patients with T less than or equal to 3 cm 5-year DFI was 54% and 5-year S was 76%, while in patients with T greater than 3 cm the respective values were 23% (p less than 0.001) and 41% (p less than 0.001). These significant DFI and S differences persisted after adjustment for number of N+ by bivariate analysis. Multivariate analysis supported the importance of T greater than 3 cm as a strong adverse predictor. Four adverse variables, T greater than 3 cm, number of N+ greater than or equal to 4, multifocality, and tumor spread in axillary fat were used to divide our patients into three subsets with significantly different DFI: Group I, with none of the above factors; Group II, with only one factor present; and Group III, with more than one factor present (5 years DFI 66%, 45%, and 21%, respectively; p less than 0.001).