This study was conducted to determine the accuracy of frozen section diagnosis in borderline ovarian tumors. Thirty-three patients were evaluated on the basis of frozen sections between February 1992 and December 1997. Frozen section diagnosis and final diagnosis were divided into three categories: for frozen section diagnosis: 1 = benign, 2 = borderline, 3 = 'at least' borderline, and for final diagnosis: 1 = benign, 2 = borderline, 3 = cancer. Three patients with a benign diagnosis according to their frozen sections were reclassified as borderline in the final diagnosis and all of them were of the mucinous type. The frozen section diagnosis of tumors of borderline malignancy was inaccurate in 3 of 23 patients. Four of 7 patients with at least borderline according to their frozen section diagnosis had invasive cancer at the final diagnosis. The correlation between frozen section diagnosis and final pathological examination was 72.7% (24/33). We found 9% (2/22) inaccurate results in the serous type and 36.6% (4/11) in the mucinous type. The sensitivity and specificity of frozen section diagnosis were found to be 86.95 and 57.14%, respectively. We concluded that frozen section evaluation in identifying a borderline ovarian malignancy is accurate enough to exclude the presence of a benign pathology. Copyright (C) 2000 S. Karger AG. Basel.