Metastatic breast cancer remains an incurable disease and the median overall survival has not significantly improved over the past two decades. Aims of the present randomized phase II trial were to analyse activity and toxicity of chemotherapies with single agent or with combination regimens in previously treated patients with advanced breast cancer. Ninety-nine eligible patients were randomized to receive the following chemotherapies: Arm A - vinorelbine 30 mg/m2 i.v. weekly; Arm B - leucovorin 100 mg/m2 i.v. followed by 5-fluorouracil 370 mg/m2 i.v. days 1 --> 5, q 28 days; Arm C - mitoxantrone 12 mg/m2 i.v. only day 1 + leucovorin 100 mg/m2 i.v. followed by 5-fluorouracil 370 mg/m2 i.v. days 1 --> 3, q 28 days. Patients characteristics are comparable in the three groups. The median number of chemotherapy courses administered was 7, 6 and 5 in arm A, B and C, respectively. Objective responses were 24%, 30% and 21% and the median duration of responses were 2, 2.5 and 5.5 months in the arm A, B and C, respectively. Median overall survivals were 9.5, 9 and 9 months in the three arms. No difference was noted comparing the survivals of responding or non responding patients. General toxicity was not mild, with 27.5% of patients experiencing WHO grade 3-4 toxicities. Our results are similar in the three groups of patients and comparable to those reported by other authors. Chemotherapy applied to patients with second or subsequent recurrence allow objective responses in a small percentage of patients. Moreover responders have a negligible prolongation of survival.