To increase the dose-intensity of two drugs in metastatic breast cancer, we tested the feasibility, in phase I studies, of two schedules of epirubicin (E) and cyclophosphamide (C) - sequential (E--> C) and alternating (E/C) - with respect to the standard combination (EC). Drugs were given at three planned-dose levels, plus either G-CSF or GM-CSF. Patients with metastatic (30), inoperable stage IIIb (2) or inflammatory (7) breast cancer were treated. The doses of EC, given every 21 days (4 cycles), were 75/1500, 82.5/2250, 90/3000 mg/m2. In the E/C schedule, epirubicin was given at cycles 1, 3 and 5, and cyclophosphamide at cycles 2, 4 and 6. In the E--> C schedule, three cycles of epirubicin then three cycles of cyclophosphamide were administered. In both experimental schedules, drugs were given every 14 days for 6 cycles at doses of 100, 110, 120 mg/m2 (E) and 2000, 3000, 4000 mg/m2 (C). The average relative dose-intensity was 1.2-fold and 2-fold greater with E/C and E--> C, respectively, than with EC. The third level dose was feasible with all schedules. Grade 4 leucopenia occurred in 77% of patients. Thrombocytopenia was absent in 6 cases and grade 4 in 12 (30.8%). Eighty-one percent of patients on experimental schedules required red blood cell support versus 44.4% of patients on EC. At the third level, platelet transfusions were more frequent among patients treated with EC (27. 8%). Non-haematological toxicity was mild: about 20% of patients experienced grade 3 vomiting, irrespective of schedule. Only 2 patients had grade 3 mucositis; no patient developed heart failure. Fever (61% of patients) and bone pain (55.5% of patients) were relevant in the GM-CSF treated groups and 12 patients shifted to G-CSF. The overall response rate was 84.6%: 5/39 (12.8%) complete response and 28/39 (71.8%) partial response. At 30/9/98, median survival was 29.5 months, with no difference between patients with metastatic and stage IIIb/inflammatory breast cancer. Median follow-up of surviving patients was 62 months (range 17-83). The 5-year estimated survival was 19% (95% confidence intervals = 7-31%). Rapidly alternating or sequential cycles of epirubicin and cyclophosphamide with CSF support is a feasible strategy that allows a higher increase of dose-intensity of the single drugs. Hospitalization and anemia were more frequent with the experimental schedules, and thrombocytopenia with the standard schedule. Overall, this intensified therapy was very active.