The aim of this study was to determine the potential mechanism of implant fracture using 3-dimensional motion analysis of patients with rheumatoid arthritis. Active flexion motion in 9 hands (34 fingers) of 6 female patients with rheumatoid arthritis who previously underwent hinged silicone metacarpophalangeal joint arthroplasty was examined using 4-dimensional computed tomography. Positions of the proximal phalanges relative to the metacarpals were quantified using a surface registration method. The deformation of the silicone implant was classified in the sagittal plane in the maximum flexion frame. The longitudinal bone axis of the proximal phalanx and the helical axis of the proximal phalanx were evaluated in 3-dimensional coordinates based on the hinge of the silicone implant. Nineteen fingers were classified into group 1, in which the silicone implant moved volarly during flexion without buckling of the distal stem. Twelve fingers were classified into group 2, in which the distal stem of the silicone implant buckled. Three fingers were classified into group 3, in which the base of the distal stem had already fractured. Quantitatively, the longitudinal bone axes of the proximal phalanges were displaced from dorsal to volar in the middle stage of flexion and migrated in the proximal direction in the late phase of flexion. The helical axes of the proximal phalanges were located on the dorsal and proximal sides of the hinge, and these tended to move in the volar and proximal directions as the metacarpophalangeal joint flexed. Volar and proximal translation of the proximal phalange was observed on 4-dimensional computed tomography. Proximal displacement of the bone axis late in flexion appears to be a contributing factor inducing implant fractures, because the pistoning motion does not allow the implant to move in the proximal direction.