Objective:We have assessed the accuracy of frozen section diagnosis and the outcomes of misdiagnosis in borderline tumors of the ovary (BTO) according to frozen section. Methods: All pathology reports with BTO in both frozen and permanent section analyses between 1994 and 2008 at Seoul St. Mary's Hospital were reviewed. Frozen section diagnosis and permanent section histology reports were compared. Logistic regression models were conducted to evaluate the correlation of patient and tumor characteristics with diagnostic accuracy. The clinical outcomes of misdiagnosis were evaluated. Results: Agreement between frozen section diagnosis and permanent histology was observed in 63 of 101 patients (62.4%). Among the 76 patients with frozen section proven BTO, under-diagnosis and over-diagnosis occurred in 8 of 76 (10.5%) and 5 of 76 patients (6.6%), respectively. Mean diameter of under-diagnosed tumor was larger than matched BTO (21.0±11.4 vs. 13.7±7.1; p=0.021). Tumor size 20 cm was determined as the optimal cut-off for under-diagnosis (50% sensitivity, 87.3% specificity). Among 8 under-diagnosed patients, no patient relapsed. Among 5 over-diagnosed patients, 2 patients < 35 years of age had fertility-preserving surgery. Conclusion: Although frozen section diagnosis is an important and reliable tool in the clinical management of patients with ovarian tumors, over-diagnosis and under-diagnosis are relatively frequent in frozen proven BTO. Surgical decision-making for BTO based on frozen section diagnosis should be done carefully, especially in large tumors.