Understanding the differences in knee kinematic descriptions is important for comparing data from different laboratories and observing small but important changes within a set of knees. The purpose of this study was to identify how differences in fixed body femoral coordinate systems affect the described tibiofemoral and patellofemoral kinematics for cadaveric knee studies with no hip present. Different methods for describing kinematics were evaluated on a set of seven cadaveric knees during walking in a dynamic knee simulator. Three anatomical landmark coordinate systems, a partial helical axis, and an experimental setup-based system were examined. The results showed that flexion-extension was insensitive to differences in the kinematic systems tested, internal-external rotation was similar for most femoral coordinate systems although there were changes in absolute position, varus-valgus was the most sensitive to variations in flexion axis direction, and anterior-posterior motion was most sensitive to femoral origin location. Femoral coordinate systems that define the sagittal plane using anatomical landmarks and locate the flexion axis perpendicular to the femur's mechanical axis in the frontal plane were typically similar and described kinematics most consistently.