There is an increasing incidence of knee pain and injury among the population, and increasing demand for higher knee function in total knee replacement designs. As a result, clinicians and implant manufacturers are interested in improving patient outcomes, and evaluation of knee mechanics is essential for better diagnosis and repair of knee pathologies. Common knee pathologies include osteoarthritis (degradation of the articulating surfaces), patellofemoral pain, and cruciate ligament injury and/or rupture. The complex behavior of knee motion presents unique challenges in the diagnosis of knee pathology and restoration of healthy knee function. Quantifying knee mechanics is essential for developing successful rehabilitation therapies and surgical treatments. Researchers have used in-vitro and in-vivo experiments to quantify joint kinematics and loading, but experiments can be costly and time-intensive, and contact and ligament mechanics can be difficult to measure directly. Computational modeling can complement experimental studies by providing cost-effective solutions for quantifying joint and soft tissue forces. Musculoskeletal models have been used to measure whole-body motion, and predict joint and muscle forces, but these models can lack detail and accuracy at the joint-level. Finite element modeling provides accurate solutions of the internal stress/strain behavior of bone and soft tissue using subject-specific geometry and complex contact and material representations. While previous FE modeling has been used to simulate injury and repair, models are commonly based on literature description or average knee behavior. The research presented in this dissertation focused on developing subject-specific representations of the TF and PF joints including calibration and validation to experimental data for healthy, pathological, and implanted knee conditions. A combination of in-vitro experiment and modeling was used to compare healthy and cruciate-deficient joint mechanics, and develop subject-specific computational representations. Insight from in-vitro testing supported in-vivo simulations of healthy and implanted subjects, in which PF mechanics were compared between two common patellar component designs and the impact of cruciate ligament variability on joint kinematics and loads was assessed. The suite of computational models developed in this dissertation can be used to investigate knee pathologies to better inform clinicians on the mechanisms surrounding injury, support the diagnosis of at-risk patients, explore rehabilitation and surgical techniques for repair, and support decisionmaking for new innovative implant designs.