This study performed the forensic anthropological sex estimation of Koreans in a non-metric way by reconstructing three-dimensional (3D) computed tomography (CT) images of skulls. The skull CT images used in this study were 100 (51 males, 49 females), and all CT images were taken with a slice thickness of 0.75 mm and then reconstructed into 3D images using the MIMICS 23.0 program. Using the reconstructed 3D image, measurements were repeated twice. The sex determination was male if the 4 point to 5 point was relatively more in five landmarks, and female if the points of 1 to 2 were relatively more. Results of the study show that, 88 of the 100 cases matched the actual sex. Among the 12 discrepant cases, ten cases were mismatched with the actual sex even though the estimation and repeated estimation readout of sexestimating were the same. Two cases, were “unknown,” showing different sexes in the first and repeated estimations. In conclusion, this study indicated that a forensic anthropological analysis from 3D images provided accurate point information on the landmarks of skulls, showing as high an accuracy as the sex estimation method using real bones. The ten cases of sex mismatch, except the two “Unknown” cases, are considered to be errors that did not consider differences in population groups. In further studies, further establishing a nonmetric, specifically Korean methods to increase the accuracy and reliability of sex estimation is need.