Purpose: To establish a fully automated image processing protocol for investigating in-vivo kinematics of the TMC joint using 4D-CT. Methods or Backround: 4D-CT images of fifteen healthy volunteers (7 females, 8 males) were obtained during cyclic opposition reposition motion of the thumb. Tube rotation time was 0.28 s, tube voltage 80 kVp, tube current 50 mA, 12 cm collimation, CTDIvol was 13 mGy, DLP of 156 mGycm and scan duration of 6s. Segmentation of the trapezium (Tz), first metacarpal (MC1) and second metacarpal (MC2) were obtained using a multi-atlas segmentation framework. Anatomical landmarks automatically propagated onto each of the bones were used in the creation of bone embedded reference frames. A dynamic sequential rigid registration resulted in transformations from which cardan angles were computed as kinematic parameters. Proximity maps were also estimated as a surrogate for joint contact area. Results or Findings: The predominant rotation of MC1 relative to Tz was around the flexion/extension axis 41.45°[36.63-46.26]. MC1 relative to MC2 showed predominant motion about the internal/external rotation axis of MC2 50.22° [45.11 - 55.33]. The average rotation between Tz and MC2 was below 5° across all axes. A gradual decrease in proximity areas was observed from reposition towards opposition, an indication that joint is more stable at beginning of reposition. Conclusion: Our automated workflow successfully allowed the computation of TMC kinematics with significant time gain by minimising laborious userinteractions. In term, our proposed workflow and results obtained have the potential to be used as a tool for improving knowledge on TMC joint kinematics.