Abstract Relatively high rates of loosening and implant failure have been reported after total ankle arthroplasty. Abnormal kinematics and incongruency of the articular surface may cause increased contact pressure and rotational torque applied to the implant, leading to loosening and implant failure. We measured in vivo kinematics of two-component total ankle arthroplasty (TNK ankle), and assessed congruency of the articular surface during the stance phase of gait. Eighteen ankles of 15 patients with a mean age of 75±6 years (mean±standard deviation) and follow-up of 44±38 months were enrolled. Lateral fluoroscopic images were taken during the stance phase of gait. 3D–2D model-image registration was performed using the fluoroscopic image and the implant models, and three-dimensional kinematics of the implant and incongruency of the articular surface were determined. The mean ranges of motion were 11.1±4.6°, 0.8±0.4°, and 2.6±1.5° for dorsi-/plantarflexion, inversion/eversion, and internal/external rotation, respectively. At least one type of incongruency of the articular surface occurred in eight of 18 ankles, including anterior hinging in one ankle, medial or lateral lift off in four ankles, and excessive axial rotation in five ankles. Among the four ankles in which lift off occurred during gait, only one ankle showed lift off in the static weightbearing radiograph. Our observations will provide useful data against which kinematics of other implant designs, such as three-component total ankle arthroplasty, can be compared. Our results also showed that evaluation of lift off in the standard weightbearing radiograph may not predict its occurrence during gait.