Pelvic resections to remove bone tumors can lead to large bone defects, which vary based on the location and extent of the tumor. Recent advancements in 3D-printing of biocompatible alloys, along with high-resolution computed tomography (CT) imaging, have made custom implants a viable option for reconstructing and restoring load-bearing ability and walking function in individuals with pelvic sarcoma. However, prosthetic re-construction of the pelvis has been associated with high complication rates and poor long-term functional outcomes. Aseptic loosening is one of the leading causes of post-surgery complications. A significant contributor to implant loosening is bone resorption due to stress-shielding. This phenomenon occurs in peri-prosthetic bone when the stiffer implant shields the less stiff bone from the mechanical stimuli required for maintaining healthy bone density. Incorporating porosities in implants using 3D-printable lattices can reduce their stiffness, hence abating the stress-shielding. Moreover, implant fixation to bone can be improved by promoting bony ingrowth within the implant. This thesis presents a computational framework for designing custom pelvic im-plants with enhanced osseointegration and stress-shielding resistance. Computational models, capable of predicting bone stresses after surgery, were needed for evaluating stress-shielding. First, patient-specific finite element models of bone with heterogeneous properties were developed. Different bone material mapping methods were investigated. The most suitable methods in terms of model convergence were identified. The long-term fixation of the implant to bone through osseointegration was also addressed. Second, a computational approach was outlined for evaluating the performance of implant lattices for bone ingrowth and matching bone properties. Superior biocompatibility of the im-plant with the peri-prosthetic bone was achieved by tuning the anisotropic elastic proper-ties of the implant to match those of bone. We used a previously validated model for pre-dicting bone growth inside porous implants. Thus, lattice designs with balanced osseoin-tegration and bone material properties matching performance were realized. Finally, a topology optimization formulation was developed for designing stress-shielding resistant pelvic implants. This formulation yielded a design that significantly reduced the predicted volume of bone resorption at the bone-implant interface compared to the solid implant. Therefore, a pelvic implant design with the potential for improved long-term stability was obtained.