The relationship between altered tibiotalar and subtalar kinematics and development of ankle osteoarthritis is unknown, as skin marker motion analysis cannot measure articulations of each joint independently. Here, we quantified the accuracy and demonstrated the feasibility of high-speed dual fluoroscopy (DF) to measure and visualize the three-dimensional articulation (i.e., arthrokinematics) of the tibiotalar and subtalar joints. Metal beads were implanted in the tibia, talus and calcaneus of two cadavers. Three-dimensional surface models of the cadaver and volunteer bones were reconstructed from computed tomography images. A custom DF system was positioned adjacent to an instrumented treadmill. DF images of the cadavers were acquired during maximal rotation about three axes (dorsal-plantar flexion, inversion-eversion, internal-external rotation) and simulated gait (treadmill at 0.5 and 1.0 m/s). Positions of implanted beads were tracked using dynamic radiostereometric analysis (DRSA). Bead locations were also calculated using model-based markerless tracking (MBT) and compared, along with joint angles and translations, to DRSA results. The mean positional difference between DRSA and MBT for all frames defined bias; standard deviation of the difference defined precision. The volunteer was imaged with DF during treadmill gait. From these movements, joint kinematics and tibiotalar and subtalar bone-to-bone distance were calculated. The mean positional and rotational bias (±standard deviation) of MBT was 0.03±0.35 mm and 0.25±0.81°, respectively. Mean translational and rotational precision was 0.30±0.12 mm and 0.63±0.28°, respectively. With excellent measurement accuracy, DF and MBT may elucidate the kinematic pathways responsible for osteoarthritis of the tibiotalar and subtalar joints in living subjects.