Background: Ovarian cancer is the second most common type of female genital tract malignancy. Treatment planning differs for benign, borderline, and malignant subtypes of surface epithelial tumors and depends on accurate histopathological diagnosis. The aim of this study is to determine the accuracy and causes for error in intraoperative diagnosis of ovarian surface epithelial tumors.Methods: In this retrospective study, we analyzed all cases of ovarian surface epithelial tumors referred to the Pathology Department of our hospital from April 2010 to December 2015. We considered the final diagnosis as the gold standard and determined the accuracy, sensitivity, specificity, positive predictive value, negative predictive value, underdiagnosis and overdiagnosis for each group of benign, borderline, and malignant tumors. An expert pathologist blinded to the diagnosis reviewed patients’ frozen and permanent slides and categorized causes of error into misinterpretation, sampling, and technical errors.Results: We assessed 220 patients’ slides (96 benign, 66 borderline, and 58 malignant tumors). The accuracy of the frozen section was: 98% in benign, 80.3% in borderline and 67.2% in malignant tumors. The frozen sections had a sensitivity of 97.9% for benign tumors and a sensitivity of 67.2% and specificity of 100% for malignant tumors. Borderline tumors had a sensitivity of 91% and specificity of 88.4%. Mucinous borderline tumors comprised the more frequent uncertain and underdiagnosed cases. The main cause for error in this group was sampling error. In malignant neoplasms, 15.5% were reported to be at least borderline. Technical issues were the cause of difficulty in interpretation. In the benign category, cystadenofibroma could be misinterpreted as a borderline malignancy.Conclusion: Frozen section is an accurate, specific method for diagnosis of benign and malignant tumors. In the borderline category, the results should be interpreted with caution.