Objective To develop a musculoskeletal multi-body dynamic model of the patient-specific total knee replacement (TKR), and to simulate knee joint biomechanical characters of the patient during right-turn gait. Methods Based on the musculoskeletal dynamic software AnyBody and the method of force-dependent kinematics as well as the related data from a patient with TKR, the corresponding patient specific lower extremity musculoskeletal multi-body dynamic model was constructed and then used to simulate the right-turn gait of the patient. The knee contact forces, motion, muscle activations and ligament forces were predicted simultaneously by inverse dynamics analysis on such right-turn gait. ResultsThe root mean square error of the predicted average tibiofemoral medial contact force and lateral contact force were 285 N and 164 N, respectively, and the correlation coefficients were 0.95 and 0.61, respectively. The predicted average patellar contact force was 250 N. The predicted contact forces and muscle activations were consistent with those in vivo measurements obtained from the patient. In addition, the model also predicted the average range of tibiofemoral rotations of flexion-extension, internal-external, varus-valgus as 3°-47°, -3.4°-1.5°, 0.2°--1.5°, and the average range of tibiofemoral translations of anterior-posterior, inferior-superior, medial-lateral as 2.6-9 mm, 1.6-3.2 mm, 4.2-5.2 mm, respectively. The predicted average peak value of the medial, lateral collateral ligament force and posterior cruciate ligament force were 190, 108, 108 N, respectively. Conclusions The developed model can predict in vivo knee joint biomechanics, which offers a robust computational platform for future study on the failure mechanisms of knee prosthesis in clinic.