We investigated the feasibility of estimating living stature in Japanese subjects using femoral length and pelvic dimensions measured on three-dimensional (3D) pelvic models reconstructed from cross-sectional computed tomography (CT) images. For this cross-sectional study, we recruited 106 healthy Japanese subjects. Maximum and bicondylar femoral length, as well as pelvic width, depth, and height, were measured on 3D bone models reconstructed from multi-slice CT images. The correlation of stature with each parameter was evaluated, and multiple regression equations were derived as formulae for living stature estimation. Prediction accuracy was evaluated as the mean absolute difference (MAD) between the measured and estimated statures. Maximum and bicondylar femoral lengths were similar and showed strong correlations with stature (> 0.8 in both males and females). Among the pelvic dimensions, height (craniocaudal length) showed the strongest correlation with stature in both males (r = 0.649) and females (r = 0.684). Formulae using femoral length plus pelvic height provided the best estimation of living stature in both males and females (MAD, 25-26 mm). Among the studied pelvic dimensions, height provided the best estimation of living stature when used alone (MAD, 34-36 mm) in both males and females. The intraclass correlation coefficients were high (> 0.9) for both intraobserver and interobserver reliability. Femoral length and pelvic height measured on CT images are reliable predictors of living stature in the Japanese population. Such tools are particularly useful in disaster victim identification, when the long bones are often not intact but the pelvic bones are.