In the design of femoral stem prostheses, both biologic factors and prosthetic design features influence the long-term performance of the implant. The analysis of all these influences is difficult, if not impossible, at the present time. This research presents a computational approach, using automatic contour detection, octree automesh and a special finite element program, that allows the stress analysis of different prostheses to be performed with minimal human interaction. Using computed tomography scans of a femur, the stress pattern in the femur when subjected to physiologic loading with or without a implanted prosthesis can be generated. A variety of prosthetic designs can be evaluated and the most suitable design can be selected for a particular patient prior to the operation. With an accurate inner contour generated by the computer algorithm developed in this study, a femoral prosthesis can be designed which most closely reproduces the stress pattern in the intact femur. Based on this study, different design parameters can be evaluated and the optimal prosthetic design can then be carried out with more advanced computer hardware.