Osseointegrated trans-femoral implants are a promising orthopaedic anchoring method to attach prosthetic limbs. The clinical success of this technique depends on the biological osseointegration achieved between the implant and bone. In this thesis, resonant characteristics of the trans-femoral osseointegration implant systems were investigated thoroughly by means of natural frequency (NF) analysis. The study consists of three parts, Finite Element (FE) simulation, in-vitro and in-vivo experiments. In the FE analysis study, various clinical conditions were simulated, such as, size of contact area, mechanical properties of the interface, femur length, and mechanical properties of soft tissue. The results showed that both the area and elastic modulus of the osseointegration interface affect the NF of the implant system. The in-vitro experiments were carried out using a silicone interface layer to simulate an implant-bone interface. The results showed a linear correlation between the NF and the elastic modulus of the interface. Meanwhile, the NF was found decreased when the length of the abutment increased. The in vivo experiments were carried out on three patients during their rehabilitation programme. Gradual NF increases were found with patients who finished the rehabilitation program within normal time. NF decreases were commonly observed after the first weight bearing. Moreover, all of the NF results after the weight bearing were significant lower than the NF results before the weight bearing (p<0.01). The correlation found between the NF and mechanical characteristics, such as contact area and elastic modulus, of osseointegration interface in all of above three studies suggested that the NF analysis is an effective method to reveal the osseointegration condition during the rehabilitation. As the future work the author suggested further in vivo investigations with established techniques including radiography and histology would help to confirm the correlation uncovered in study. The suggestion of further improvement of the NF analysis software and FE models were also given.