We aimed to reproduce the anthropometrical measurement of femoral dimensions using multi-planar reconstruction computed tomography (CT), assess the correlation between stature and femoral measurements obtained by this approach, and establish a regression equation for estimating stature in the modern Japanese population. We used data regarding 224 cadavers (116 males, 108 females) that were subjected to postmortem CT and subsequent forensic autopsy at our department between October 2009 and July 2016. To simulate the placement of the femur on the osteometric board using reconstructed CT images, we defined a virtual horizontal plane (VHP) based on the three most dorsal points of the femur (lateral condyle, medial condyle, and greater trochanter). Five femoral measurements including the maximum femoral length (MFL) were obtained. The correlations between stature and each femoral measurement were expressed in terms of the coefficient of determination (R). On regression analysis, MFL provided the lowest value for the standard error of the estimation (SEE); the SEE values in all subjects, males, and females, respectively, were 3.783 cm (R = 0.832), 3.850 cm (R = 0.653), and 3.340 cm (R = 0.760) for MFL on the left side and 3.747 cm (R = 0.835), 3.847 cm (R = 0.650), and 3.290 cm (R = 0.687) for MFL on the right side. Multiple regression equations using MFL and femoral epicondylar breadth were slightly superior to simple regression equations in males and in all subjects (SEE = 3.44-3.55 cm), whereas no effective equation could be obtained in females. To our knowledge, this is the first multiple regression equation for stature estimation using only femoral measurements.