To determine the accuracy of frozen section according to the status of malignancy and the histologic cell type, we reviewed the frozen and permanent pathologic reports of 212 resected ovarian masses in our hospital. The accuracy, sensitivity, specificity, positive, and negative predictive value of frozen section were studied. The overall accuracy to determine the status of malignancy was 90.9%. Sensitivity of the test was highest in the benign groups at 99.1% and lowest in the borderline groups at 50%. All inaccurate diagnoses were in the common epithelial groups. Problems in diagnosis of mucinous tumors and borderline tumors were striking. The accuracy of the test for histologic diagnosis was 91.9%. Most cases of the incorrect diagnosis (81.3%) were common epithelial tumors. In conclusion, the accuracy of frozen section in the diagnosis of ovarian mass was generally high with a few exceptions in large tumors, mucinous, or borderline tumors that yielded lower accuracy, sensitivity, specificity, and positive predictive value. We encourage both the surgeons and the pathologists to be cautious of these limitations. Additional number of frozen section taken for a mass larger than 10 cm may minimize the error in large tumors to some extent.