Purpose: To evaluate the accuracy of a frozen section and to analyze the factors affecting frozen section results in cases of borderline ovarian tumors (BOTs). Materials and Methods: The files and pathological reports of 82 cases diagnosed with BOT at our clinic, between January 1994 and June 2009, have been retrospectively evaluated. The frozen section results were compared to the permanent paraffin section results. Accuracy, overdiagnosis, and underdiagnosis rates were estimated. The factors affecting the diagnosis were also evaluated using logistic regression analysis. Results: The mean age was 40.16 14.01 years. Of the patients, 47.6 had serous and 42.7 had mucinous histology. About 90 of the cases were in stage I. The rate of correct diagnosis with frozen section was 69.5. The rates of overdiagnosis and underdiagnosis were 1.2 and 29.3; respectively. The factors affecting the diagnosis were determined as, the dimension of the ovarian mass (P = 0.005), presence of a solid component (P = 0.000), preoperative CA 125 value (P = 0.016), and intraoperative rupture of the ovarian cyst (P = 0.049). Conclusion: In the frozen section evaluation of BOTs, the underdiagnosis that restricts the diagnostic performance of the method seems to be a major problem. A more careful approach is therefore needed, while choosing a proper surgical technique during laparotomy for ovarian masses. In order to reduce the false diagnosis and surgical morbidity, the frozen section analysis should be applied by experienced pathologists and the possible predictive factors affecting a false diagnosis should carefully be taken into consideration.