Between 1977 and 1994, our center administered successively 4 different chemotherapy regimens to 242 evaluable patients with locally advanced breast cancer. Patients with inflammatory signs were excluded. Sixty-eight patients were treated by AVCF (A (adriamycine) + V (vincristine) + C (cytoxan) + F (5FU)), 47 by AECF (A + E (vindesine) + C + F), 81 by CAFP (C + A + F + P (prednisone)) and 46 by AN (A + N (vinorelbine)). The mean number of cycle was 3. One hundred and twenty-five patients (52.5%) responded to chemotherapy and we recorded 35 complete response (14.7%). The response rates at the different combinations were respectively: AVCF: 29.4%, AECF: 53.2%, CAFP: 64.9%, AN: 65.2%, and were independent of tumor size, grade and receptor status. The response rate at the AVCF regimen was significantly worse than the others (p = 0.0005). Breast conserving surgery was performed in 31 patients (14%) and 17 patients (8%) had a complete response. Among the 35 patients with complete response, 21 were treated by radiotherapy alone. Local recurrence occurred in 19 patients (7.9%) and 96 (40%) had advanced disease. The mean follow-up of AVCF regimen was 150 months, 115 months for AECF, 111 for CAFP and 42 months for AN. The disease-free survival and the overall survival were significantly better with AECF, CAFP and AN regimens (DFS p < 0.04, OS p < 0.02). Survival was better in those patients with an objective response (p = 0.002) or with non-affected axillary node at the time of surgery. Our study showed already that AVCF combination was significantly lower than AECF, CAFP, AN in terms of response rate, disease-free survival and overall survival. Waiting the results of randomized studies about the impact of neoadjuvant chemotherapy on survival, we look for chemotherapy regimen improving the rate of conservative surgery.