<i>Objective:</i> Breast cancer patients with >3 involved nodes (N+) have a poor outcome. Chemotherapy (CT), alone or combined with endocrine therapy (ET) in hormone receptor (HOR)-positive patients, is the standard for these women. However, there are still questions surrounding the optimal adjuvant CT regimen. <i>Methods:</i> 244 patients with >3 N+ were randomized to receive either four 3-weekly courses of epirubicin (E: 100 mg/m², day 1) followed by four 4-weekly cycles of cyclophosphamide, methotrexate and 5-fluorouracil (CMF: 600, 40, 600 mg/m², days 1, 8: n = 122) or four 3-weekly courses of paclitaxel (T: 175 mg/m², day 1) followed by four 3-weekly cycles of epirubicin and vinorelbine (E: 75 mg/m², day 1; V: 25 mg/m², days 1, 8: n = 122). After CT, tamoxifen (plus an LH-RH analog in menstruating women) was given to all HOR-positive patients over a period of 5 years. Overall survival (OS) was the primary end point. Relapse-free survival (RFS) and toxicity were secondary end points. <i>Results:</i> At a median follow-up time of 102 months (range 3–146), OS and RFS did not differ significantly between groups (E-CMF vs. T-EV: OS, HR 0.94, 95% CI 0.59–1.48, p = 0.8; RFS, HR 0.86, 95% CI 0.57–1.29, p = 0.45). The lack of any difference between assigned treatments was confirmed by multivariate analysis (E-CMF vs. T-EV: RFS, HR 0.98, 95% CI 0.64–1.48, p = 0.9). The 2 regimens showed different toxicity profiles. In fact, significantly more women assigned to E-CMF were affected by stomatitis (p = 0.001) while significantly more women in the T-EV group developed peripheral neuropathy (p < 0.0001) and musculoskeletal disorders (p < 0.0001). However, side effects were moderate and manageable and no toxic death occurred in either arm of the study. <i>Conclusions:</i> T-EV was safe and moderately toxic but was not superior to E-CMF.