We evaluated the contribution of an epirubicin-based adjuvant chemotherapy on disease-free survival (DFS) in poor prognosis, node-negative breast cancer (BC) patients. Poor prognostic factors were defined as: pathologic tumor size >or= 4 cm, estrogen-receptor negative, and progesterone-receptor negative. Scarff-Bloom Richardson grade 2 tumors must have two of these factors, and only one in case of grade 3. Between 1988 and 1994, 328 patients were randomized to receive either no systemic treatment (control, n = 161), or fluorouracil 500 mg/m(2), epirubicin 50 mg/m(2), cyclophosphamide 500 mg/m(2), 6 cycles every 21 days (FEC50, n = 167), without any hormonal treatment. The median follow up was 114 months. The 10-year DFS rates were 64 and 71%, respectively (p = 0.23). In the Cox regression model, independent prognostic factors of relapse were the number of nodes examined < 10 (p = 0.002), BCS (p = 0.01), and premenopausal status (p = 0.04). In this model, the relative risk of relapse was 1.46 (CI95 %: 1.05-1.87) in favor of FEC50. In patients who underwent BCS, 21 % developed a local relapse (24 versus 18 %, respectively). The 10-year local DFS was 70.5 and 79.3 %, respectively (p = 0.27). The 10-year overall survival was not different (74.1 versus 70.7 %, p = 0.82). After 10 years of follow-up, the FEC50 regimen reduced the risk of relapse in poor-prognosis node-negative BC patients. The incidence of local relapse was high, and probably related to inclusion criteria. Epirubicin was probably underdosed in such patients, and ongoing studies using 100 mg/m(2) of epirubicin will give us the answer in a near future.