In a laboratory study using seven fresh-frozen anatomic specimen knees, the effect of total knee arthroplasty on the three-dimensional kinematics of the patella, femur, and tibia were measured. Experiments were performed in the intact knee, after division of the anterior cruciate ligament (ACL), after total knee arthroplasty, and after 10 degrees internal rotation, 10 degrees external rotation, 5-mm medial shift, and 5-mm lateral shift of the femoral component on the femur. The presence of a high lateral ridge on the anterior surface of the femoral component effectively prevented patellar subluxation or dislocation, but displaced and tilted the patella medially. Internal rotation or medial displacement of the femoral component exaggerated this medial patellar displacement and shift. External rotation of the femoral component corrected it, except at flexion angles greater than 100 degrees, where the femur was shifted medially on the tibia and externally rotated 15 degrees. This combination produced a net 10-mm medial displacement of the patella relative to the tibia at 120 degrees knee flexion. Lateral placement of the femoral component compensated for the effect of the high lateral ridge and allowed more normal patellar tracking while allowing tibiofemoral motions similar to those seen after sectioning of the ACL. The kinematics of the patellofemoral and tibiofemoral joints were not reproduced with a total knee prosthesis that sacrifices the ACL. When using a prosthesis with a high lateral ridge, lateral placement of a femoral component prevented patellar dislocation and allowed patellar tracking patterns similar to those seen in the intact knee without further altering tibiofemoral motions.