Fifty-five consecutive patients with metastatic breast cancer (MBC) (n = 57) were treated with a combination of levofolinic acid (I-FA) 100 mg/m2 plus 5-fluorouracil (5-FU) 340 mg/m2 i.v. on day 1-3, cyclophosphamide (CTX) 600 mg/m2 i.v. on day 1 and mitoxantrone (DHAD) 12 mg/m2 i.v. on day 1. DHAD dose was progressively escalated by 2 mg/m2/cycle up to 18 mg/m2 in the absence of dose-limiting toxicities. Granulocyte colony stimulating factor (G-CSF) was given s.c. in order to prevent neutropenia. DHAD dosage could be increased to 18 mg/m2 in 66 out of 317 cycles of chemotherapy (21%). In most patients the dose-limiting toxicity was represented by myelosuppression. A statistically significant correlation was found between median white blood cell (WBC) or absolute neutrophil count (ANC) nadir and DHAD dose level. Moreover, a statistically significant correlation was observed between the number of chemotherapeutic cycles, nadir ANC and WBC, and the occurrence of anemia and thrombocytopenia of increasing severity. These data suggest the occurrence of progressive cumulative bone marrow toxicity. Although patients who reached different DHAD levels showed differences in mean dose intensity, such differences were not statistically significant. No correlation was found between the increase in dose intensity and type, rate or duration of objective responses. In patients with metastatic breast cancer the overall response rate was 72% (95% CL 57-84%) with a 18% complete response rate. Median duration of response was 12 and 11 months, respectively, for complete and partial responses. Projected median survival of the whole series of patients with MBC was 18 months. These data demonstrate that the combination of 5-FU with I-FA, CTX and DHAD is very active against MBC. G-CSF use allows the increase DHAD dosage up to 18 mg/m2/cycle, but its use may be linked to the occurrence of sometimes severe cumulative hematological toxicity.