Weekly chemotherapy administration represents an emerging option for the treatment of metastatic breast cancer. In order to identify clinical and biological prognostic factors for outcome, we performed a multivariate analysis in a 10-year experience of weekly chemotherapy for metastatic breast cancer patients. The original databases of phase II trials of metastatic breast cancer patients who had undergone first-line weekly chemotherapy were collected. Clinical and biological covariables were screened for a possible relationship with time to progression and overall survival in a Cox model. From 1990 to 2003, 184 patients were enrolled in three consecutive phase II studies, to evaluate activity and tolerability of weekly epirubicin with lonidamine or vinorelbine or paclitaxel. All patients were evaluable for clinical variables; histological samples were available in 40 patients. At a median follow-up of 24 months, median time to progression was 9 months (95% confidence interval 8-10) and median overall survival was 34 months (95% confidence interval 24-42). Independent variables were response (hazard ratio 2.34, P<0.0001), receptor status (hazard ratio 1.62, P=0.01) and performance status (hazard ratio 2.31, P<0.0001) for time to progression, and response (hazard ratio 1.86, P=0.005), performance status (hazard ratio 2.81, P<0.0001), dominant metastatic site (hazard ratio 2.27, P<0.0001) and enrollment period (hazard ratio 2.51, P=0.001) for overall survival. Although no biological factors were entered into the Cox model owing to the small sample size, some subpopulations showed a negative trend in survival. In our series of patients who had undergone weekly chemotherapy for metastatic breast cancer, independent prognostic factors for survival improvement were responders, performance status 0-1, nonvisceral dominant metastatic site and enrollment period. A greater sample population is needed to extensively screen for biological prognostic factors.