Direct anchoring of a lower-limb prosthesis to the bone (osseointegration) has been shown to solve some common problems associated with conventional socket prostheses. During the rehabilitation phase, amputees fitted with osseointegrated implants apply static loading against the abutment on a weighing scale to prepare the bone to tolerate the forces likely to be experienced during walking. However, the weighing scale measures only the vertical force. Moments and other directions of forces, which can affect the bone-implant interface stresses and the rehabilitation outcome, are not measured. When the amputee starts to walk, in addition, there is a risk of bone mechanical failure. This chapter illustrates the development of a finite element (FE) model to study the stresses in the bone and at the bone-implant interface. Bone-implant interface stresses were compared under three loading conditions: (1) vertical force corresponding to the load clinically prescribed in a weight-bearing exercise; (2) loads applied on the three axes, corresponding to the "true" load measured by a triaxial load transducer during the same exercise; and (3) loads experienced during independent level walking. An additional parametric analysis was performed to predict the bone structural integrity under different conditions. The model revealed that the weighing scale in fact applied much greater and less uniform stresses on the bone than expected. During walking, high stress occurred at a location different from that experienced during the weight-bearing exercise. In addition, with a bone loss of 40% and the application of a push-off loading at four times its normal force, bone failure would occur. These findings imply that triaxial loading should be monitored during weight-bearing exercises and carefully prescribed.