This retrospective multicentric study assessed the prognostic value of the serum CA 125 assay and the common clinicopathological variables in 225 patients with advanced epithelial ovarian cancer. All of these patients had serum CA 125 > or = 35 U/ml before the first cycle of chemotherapy and had had serial antigen determinations during early chemotherapy. By univariate analysis pathological complete response rate was significantly related to stage, size of residual disease after first surgery, serum CA 125 before the first cycle and before the third cycle of chemotherapy, and serum CA 125 half-life. Multiple logistic regression showed that residual disease (P = 0.002), serum CA 125 half-life (P = 0.004), serum CA 125 before the third cycle (P = 0.004), and serum CA 125 before the first cycle (P = 0.03) retained a significant value in predicting second-look findings. By log-rank test survival was significantly related to stage, residual disease, tumor grade, serum CA 125 before the third cycle, and serum CA 125 half-life. Cox proportional hazard model showed that residual disease (P = 0.0001), serum CA 125 half-life (P = 0.007), and tumor grade (P = 0.01) retained a significant value in predicting survival. In conclusion, serum CA 125 half-life during early chemotherapy was an independent prognostic factor for both the achievement of a pathological complete response and the survival of patients with advanced epithelial ovarian cancer.