Total joint replacements are highly successful in relieving pain and restoring movement of damaged joints. However, the lifespan of the implants is limited. The implant's long-term stability depends largely on the preservation of periprosthetic bone. Debris-wear particulates were first identified as the factor inducing periprosthetic bone loss. However, it was shown later that the resorption process starts before the particulates reach the periprosthetic bone. Thus a mechanical factor, interface micromotions, has been suspected to be the initiator of early bone loss. In the first part of this thesis, we investigated the response of bone cells to micromotions. Using an ex vivo setup, we applied micromotions on fresh human bone cores and showed that micromotions could indirectly activate osteoclasts after only 1 hour. Thus micromotion-related osteoclastic activity could be the initiator of periprosthetic bone loss. Local release of bisphosphonate seemed therefore to be an ideal solution to prevent early bone loss, as the target is a local process. The few previous studies assessing local release of bisphosphonate in vivo reported increased periprosthetic bone stock and/or strengthen implant fixation. But these studies only concerned rats and were designed empirically. In the second part of the thesis, to overcome the limitation resulting from the sole usage of rats, we performed a study of local zoledronate release in sheep trabecular bone. We measured that, at 4 weeks post-operatively, periprosthetic bone was 50% denser with zoledronate than without. To further analyze these experimental results, a theoretical model of bone adaptation with bisphosphonate was developed. This new model updated an existing mechanically-driven bone adaptation model by adding an equation of drug-driven bone adaptation. The identification of the model parameters was carried-out with the existing experimental data. Based on the new model, the dose of zoledronate that would maximize the periprosthetic bone density was calculated; this dose was then tested in vivo and the results validated the model's prediction with a good accuracy. The fundamental assumption in the theoretical model was that the mechanical stimulus and the drug stimulus were independent. The last aim of the thesis was then to verify this assumption. We investigated potential interactions between mechanical effects and bisphosphonate on bone adaptation in a model of adaptation of the mouse tibia to axial compression, combined with systemic injections of zoledronate. We observed that the effects of each stimulus were independent in general, however in very high strain conditions, zoledronate seemed to reduce the bone response to mechanical stimulation. On a clinical perspective, the major results of the thesis were, first, that local micromotions can initiate periprosthetic bone resorption rapidly and thus favor aseptic loosening on long-term. This new data provides a strong rationale for the local use of bisphosphonate rather than systemic. The second major result was the demonstration that local release of zoledronate increases periprosthetic bone stock in sheep. Finally, the new theoretical model is a convenient tool to plan clinical studies and compare experimental results.