For amputated patients, direct attachment of upper leg prosthesis to the skeletal system by a percutaneous implant is an alternative solution to the traditional socket fixation. Currently available implants, the OPRA system (Integrum AB, Göteborg, Sweden) and the ISP Endo/Exo prosthesis (ESKA Implants AG, Lübeck, Germany) [1-2] allow overcoming common soft tissue problems of conventional socket fixation and provide better control of the prosthetic limb [3], higher mobility and comfort [2, 4]. However, restraining issues such as soft-tissue infections, peri-prosthetic bone fractures [3, 5–8] and considerable bone loss around the stem [9], which might lead to implant's loosening, are present. Finally, a long a residual limb is required for implant fitting. In order to overcome the limiting biomechanical issues of the current designs, a new concept of the direct intramedullary fixation was developed. The aim was to restore the natural load transfer in the femur and allow implantations in short femur remnants (Figure 1). We hypothesize that the new design will reduce the peri-prosthetic bone failure risk and adverse bone remodeling. Generic CT-based finite element models of an intact femoral bone and amputated bones implanted with 3 analyzed implants were created for the study. Models were loaded with two loading cases from a normal walking obtained from the experimental measurements with the OPRA device [10-11]. Periprosthetic bone failure risk was evaluated by considering the von Mises stress criterion [12-14]. Subsequently the strain adaptive bone remodeling theory was used to predict long-term changes in bone mineral density (BMD) around the implants. The bone mineral content (BMC) change was measured around implants and the results were visualized in the form of DXA scans. The OPRA and the ISP implants induced the high stress concentration in the proximal region decreasing in the distal direction to values below physiological levels as compared with the intact bone. The stresses around the new design were more uniformly distributed along the cortex and resembled better the intact case. Consequently, the bone failure risk was reduced as compared to the OPRA and the ISP implants. The adaptive bone remodeling simulations showed high bone resorption around distal parts of the OPRA and the ISP implants in the distal end of the femur (on average −75% ISP to −78% OPRA after 60 months). The bone remodeling simulation did not reveal any bone loss around the new design, but more bone densification was seen (Figure 2). In terms of total bone mineral content (BMC) the OPRA and the ISP implants induced only a short-term bone densification in contrast to the new design, which provoked a steady increase in the BMC over the whole analyzed period (Figure 3). In conclusion, we have seen that the new design offers much better bone maintenance and lower failure probability than the current osseointegrated trans-femoral prostheses. This positive outcome should encourage further developments of the presented concept, which in our opinion has a potential to considerably improve safety of the rehabilitation with the direct fixation implants and allow treatment of patients with short stumps