Understanding the biomechanical effect of various factors on knee behavior after anterior cruciate ligament (ACL) injury or reconstruction is instrumental for the development of an optimal surgical treatment of ACL injury that can better restore normal knee function. This paper presents the application of a three-dimensional (3D) computational knee model for parametric studies of knee kinematics in response to simulated muscle loads. The knee model was constructed using the magnetic resonance images and biomechanical experimental data of the same cadaveric human knee specimen. The kinematics of the knee predicted by the computational model was compared with that measured from different specimens in a wide range of loading conditions and flexion angles. In general, the model predictions were within the range of experimental data. The model was then used to predict knee motion, ligament forces, and contact pressure in response to a simulated quadriceps force when the knee was ACL deficient. Partial ACL injury was simulated by reducing the stiffness of the ACL in the model. The results demonstrated that even with a reduction of 75% of the ACL stiffness, the ACL still carried a significant amount of the load (more than 58%) carried by an intact ACL. The kinematics (both tibial translation and rotation) varied less than 20% compared to that of the knee with intact ACL. The 3D computational model can be a powerful tool to simulate different variables that would influence knee function after ACL reconstruction, such as the initial tension of the ACL graft, the insertion sites of the graft, multibundle grafts, graft materials, and various physiological loading conditions.