This study presents an optimized matching algorithm for a dual-orthogonal fluoroscopic image system used to determine six degrees-of-freedom total knee arthroplasty (TKA) kinematics in-vivo. The algorithm was evaluated using controlled conditions and standard geometries. Results of the validation demonstrate the algorithm’s robustness and capability of realizing a pose from a variety of initial poses. Under idealized conditions, poses of a TKA system were recreated to within 0.02±0.01 mm and 0.02±0.03 deg for the femoral component and 0.07±0.09 mm and 0.16±0.18 deg for the tibial component. By employing a standardized geometry with spheres, the translational accuracy and repeatability under actual conditions was found to be 0.01±0.06 mm. Application of the optimized matching algorithm to a TKA patient showed that the pose of in-vivo TKA components can be repeatedly located, with standard deviations less than ±0.12 mm and ±0.12 deg for the femoral component and ±0.29 mm and ±0.25 deg for the tibial component. This methodology presents a useful tool that can be readily applied to the investigation of in-vivo motion of TKA kinematics.