From December 1980 to September 1985, a total of 90 eligible patients with stage T1-3a, node-negative, and estrogen receptor-negative (less than or equal to 10 fmol/mg of cytosol protein) tumors were entered into a randomized study to assess the effectiveness of adjuvant intravenous (IV) cyclophosphamide, methotrexate, and fluorouracil (CMF) in a subset of patients at high risk of early disease relapse. High values of [3H]thymidine labeling index were documented in two thirds of 62 assessed specimens from the patient population, and one half of the patients had histologically undifferentiated tumors. Patients were allocated to either local-regional modality alone (control group, 45 women) or to CMF (45 patients) after surgery. A full dose of CMF (600 mg/m2 each of cyclophosphamide and fluorouracil, and 40 mg/m2 of methotrexate) was administered IV on day 1, and then repeated every 3 weeks for a total of 12 treatments. After a median follow-up of 80 months, the 7-year results confirmed the superiority of adjuvant CMF compared to local-regional modality alone (relapse-free survival 85% vs 42%, P = .0001; total survival 86% vs 58%, P = .006). A benefit from adjuvant CMF was observed in all subgroups, and the rates of both local-regional and distant failure were decreased. Treatment was fairly well tolerated and devoid of life-threatening toxicity. Present results confirm our previous observation concerning the dismal prognosis of node-negative and estrogen receptor-negative breast cancer patients as well as the beneficial effect of adjuvant chemotherapy in this selected subset.