Biplane roentgenography, axial roentgenography, and fluoroscopy are the usual roentgenographic methods of measuring femoral anteversion. These methods use a strict geometrical definition of anteversion. The computerized tomography method of measuring anteversion that was developed recently, and is now widely used, does not adhere to the accepted definition of anteversion and has not been tested for accuracy in a large series. In the present study, the widely used computerized-tomography method of measuring anteversion was tested on thirty-two femoral specimens. With that method, anteversion was consistently underestimated by an average of 10 degrees compared with direct measurements and was reproducible only to within +/- 3.6 degrees. Therefore, a new method of measuring anteversion using computerized tomography was developed. It was shown to be accurate to +/- 1 degree, as tested on the same specimens. This study demonstrated geometrically why the currently practiced computerized-tomography method of selecting the points that are used to define the axis of the femoral neck is not consistent with geometrical definitions of anteversion. A more accurate method for both defining the axis of the femoral neck and measuring femoral anteversion is described and recommended for clinical use.