Musculoskeletal (MS) models should be able to integrate patient-specific MS architecture and undergo thorough validation prior to their introduction into clinical practice. We present a methodology to develop subject-specific models able to simultaneously predict muscle, ligament, and knee joint contact forces along with secondary knee kinematics. The MS architecture of a generic cadaver-based model was scaled using an advanced morphing technique to the subject-specific morphology of a patient implanted with an instrumented total knee arthroplasty (TKA) available in the fifth “grand challenge competition to predict in vivo knee loads” dataset. We implemented two separate knee models, one employing traditional hinge constraints, which was solved using an inverse dynamics technique, and another one using an 11-degree-of-freedom (DOF) representation of the tibiofemoral (TF) and patellofemoral (PF) joints, which was solved using a combined inverse dynamic and quasi-static analysis, called force-dependent kinematics (FDK). TF joint forces for one gait and one right-turn trial and secondary knee kinematics for one unloaded leg-swing trial were predicted and evaluated using experimental data available in the grand challenge dataset. Total compressive TF contact forces were predicted by both hinge and FDK knee models with a root-mean-square error (RMSE) and a coefficient of determination (R2) smaller than 0.3 body weight (BW) and equal to 0.9 in the gait trial simulation and smaller than 0.4 BW and larger than 0.8 in the right-turn trial simulation, respectively. Total, medial, and lateral TF joint contact force predictions were highly similar, regardless of the type of knee model used. Medial (respectively lateral) TF forces were over- (respectively, under-) predicted with a magnitude error of M < 0.2 (respectively > −0.4) in the gait trial, and under- (respectively, over-) predicted with a magnitude error of M > −0.4 (respectively < 0.3) in the right-turn trial. Secondary knee kinematics from the unloaded leg-swing trial were overall better approximated using the FDK model (average Sprague and Geers' combined error C = 0.06) than when using a hinged knee model (C = 0.34). The proposed modeling approach allows detailed subject-specific scaling and personalization and does not contain any nonphysiological parameters. This modeling framework has potential applications in aiding the clinical decision-making in orthopedics procedures and as a tool for virtual implant design.