The utility of preoperative CA125 to predict optimal primary tumor cytoreduction in patients with advanced (stages IIIC and IV) epithelial ovarian cancer is controversial. In this paper, we retrospectively review patients with stage IIIC and IV epithelial ovarian cancer who underwent primary cytoreductive surgery from 1989 to 2001. Ninety-nine patients were identified and included in the analysis. All patients had preoperative CA125 levels measured. Operative and pathology reports were reviewed. Optimal cytoreduction was defined as largest volume of residual disease < 1 cm in maximal dimension. Mean values were compared with t-test on a log scale when needed. The optimal cut-point for discriminating between those with vs. without optimal cytoreduction was determined using the receiver operator curve (ROC) method. Optimal cytoreduction was achieved in 73% of patients. Among patients with optimal cytoreductive status the mean CA125 level was 569, while among patients with suboptimal cytoreduction the mean CA125 level was 1520 (P < 0.007). A CA125 level of 912 was identified as the optimal cut-point to distinguish the two groups. Using this CA125 level, the sensitivity of this test in predicting optimal cytoreduction was 58% and the specificity was 54%. The positive predictive value of CA125 for optimal cytoreduction was 78% and the negative predictive value was 31%. We conclude that CA125 level is a weak positive and negative predictor of optimal cytoreductive surgery in patients with advanced epithelial ovarian cancer. The CA125 level should not be used as a primary predictor of the outcome of cytoreductive surgery and should be viewed in the context of all other preoperative features.