Two hundred forty-seven patients with metastatic breast cancer entered into the controlled trial. Its aim was to define the optimal modality of association between hormonotherapy and chemotherapy. Chemotherapy was given a monthly course of an association including: adriamycin: 45 mg/m2 on day 1; cyclophosphamide: 400 mg/m2 on days 1, 2, 3; 5 fluoro-uracil: 500 mg/m2 on days 1, 2, 3; methyl-prednisolone: 80 mg/m2 on days 1, 2, 3. Hormonotherapy was tamoxifen (TAM) at the daily dose of 30 mg. 82 patients in group I were given TAM alone for 4 months and then chemotherapy + TAM; 83 patients in group II were given simultaneously TAM + chemotherapy; 82 patients in group III were given chemotherapy alone for 8 months and then TAM + chemotherapy. The response rates in groups I, II and III were respectively of 59, 74 and 62%. The difference in favour of group II was marginally significant. The survival curves were significantly higher in group II and III than in group I (P = 0.04). This result appears as the consequence of the poor prognostic of the sub-group of 45% patients who did not respond to TAM. These results seem to emphasize that target cells of hormonotherapy are not target cells of chemotherapy and that this difference is persisting for long time under treatment, that others modalities of association between chemotherapy and hormonotherapy must be studied with the aim of reducing the kinetic's implication of chronic administration of TAM.