The aim of this study was to establish standards for determining sex from fragmentary and complete femurs in a Korean population. The statistical analysis of 12 variables (6 about breadth and 6 about length) based on 100 Korean femurs (from 50 males and 50 females) showed that all variables have significant sex differences. The most accurate discriminant variable was the condylar breadth parallel with epicondylar breadth (87.6% accuracy). In contrast, the transverse shaft diameter was not a discriminant variable for sex determination (67.0% accuracy). Breadth-related variables were generally more accurate than length-related variables. Three variables (vertical diameter of the neck [VDN], medial epicondylarlength [MCL], and condylar breadth [CB]) were selected from stepwise analysis fordiscriminating sex (93.5% accuracy). The discriminating equation was as follows: 0.171 × VDN + 0.172 × MCL + 0.128 × CB2 - 21.471. The results of this study are helpful for determining sex, even if a femur is found in a fragmented condition in the field.