Background:Accurate knowledge of knee joint kinematics, especially patellofemoral joint kinematics, is essential for prosthetic evaluation so as to further improve total knee arthroplasty performances. Improving the evaluation of the functioning of the extensor apparatus appears, in this respect, particularly important in this optimization effort. Objectives:The aim of this study was to propose a new experimental setup for the analysis of knee joint kinematics and to validate its relevance in terms of accuracy and uncertainty. The technique developed herein combines 3D reconstruction imaging with the use of a motion capture system. Material and methods: Eight pairs of fresh-frozen cadaver specimens with no evidence of previous knee surgery were studied using a new test rig where the femur remains fixed and the tibia is free to rotate. The flexion--extension cycles were executed using computer-controlled traction of the quadriceps tendon combined with an antagonist force applied to the distal part of the tibia. Knee joint kinematics were tracked using an optoelectronic motion capture system after a preliminary stage of data acquisition of bone geometry and markers position. This stage was carried out using a new digital stereophotogrammetric system, EOS® , combined with specific 3D reconstruction software that also determined the coordinate system used in the kinematic analysis. The resulting uncertainty was assessed as was its impact on the estimated kinematics. Results:Test results on eight knees validated the setup designed for the analysis of knee joint kinematics during the flexion--extension cycle. More specifically, the statistical results show that measurement uncertainty for rotations and translations remains below 0.4 and 1.8 mm, respectively, for the tibia and 0.4 and 1.2 mm for the patella (±2 S.D. for all four measurements).