BRCAPRO and Myriad II are widely used models for predicting BRCA1/2 mutation probability before genetic testing. However, the accuracy of these models in Koreans is not known. This study was performed to evaluate the accuracy of the BRCAPRO and Myriad II models. Two hundred thirty-six women with breast cancer who underwent comprehensive BRCA1/2 genetic testing at our hospital between 2003 and 2010 were included in this study. We evaluated the performance of each model by comparing the numbers of observed versus predicted mutation carriers. We calculated sensitivity, specificity, and predictive values at 10 % estimated probability. Forty-six individuals were identified to carry a deleterious BRCA mutation. The prevalence of BRCA mutation (19.5 %) was significantly higher than that predicted by BRCAPRO (9.0 %, p = 0.001) and Myriad (5.6 %, p < 0.001). In familial breast cancer patients, BRCA mutation rate (observed 22.7 %) was underestimated by both BRCAPRO (expected 11.4 %, p = 0.006) and Myriad II (expected 6.4 %, p < 0.001). Subgroup analyses showed that both models underestimated the risk of BRCA mutation in patients with a family history of breast cancer (probands' age at breast cancer diagnosis >50 years), with only one relative with breast cancer, and with non-familial early-onset breast cancer or bilateral breast cancer. Using a 10 % cut-off, the sensitivities were 47.8 % (BRCAPRO) and 50.0 % (Myriad), and positive predictive values were 44.9 % (BRCAPRO) and 43.4 % (Myriad). Both BRCAPRO and Myriad II underestimated the risk of BRCA1/2 mutation in Koreans. Our findings suggest that these models are less sensitive in Korean women, and therefore a new BRCA mutation prediction model based on Korean data is needed for proper genetic counseling.