The objective of this study was to evaluate the ability of a preoperative serum CA125 to predict whether optimal debulking (OD) could be achieved for patients with stage III and IV epithelial ovarian cancer (EOC). The records of consecutive patients who underwent primary surgery for EOC at Indiana University Hospital between January 1997 and January 2003 were reviewed. Eligibility criteria included FIGO stage III/IV disease, surgery by gynecologic oncology faculty, preoperative CA125, and an operative note clearly defining volume of residual disease. The Medcalc software statistical package was used to generate a receiver-operating characteristic (ROC) curve. Two hundred and eighty-nine cases of stage III/IV EOC were identified, of which 164 met the eligibility criteria. Serum CA125 =400 was associated with OD >/=75% of the time. Conversely, OD was achieved in =40% of patients with CA125 >/=4500. The area under the ROC curve for CA125 was .670. The OD rate for those with and without ascites was 49% and 79%, respectively (P < 0.001). In a multivariate analysis using CA125, age, and ascites, the area under the curve was 0.686. We conclude that preoperative serum CA125 did not reliably predict OD in patients with stage III-IV EOC.