The specific aim of this study was to develop a quantitative method and a kinematic method to evaluate the maximal workspace of the trapeziometacarpal (TM) joint. Six fresh-frozen human cadaver hands were disarticulated 4 cm proximal to the wrist joint and used in this experiment. The three-dimensional motion data of the TM joint was collected by an electromagnetic tracking device at 30 Hz. The workspace was reconstructed according to a complete set of motion data included circumduction, flexion-extension and abduction-adduction. A spherical fitting technique was used to obtain a sphere encompassing all the motion trajectories and estimating the centre of the sphere. The surface area of the maximal TM workspace, located on the one part of the sphere surface, was calculated by surface integration. The interclass correlation coefficient values for the reliability estimation of the repeated measurements of the radius and surface area of all specimens were 0.91 and 0.98 respectively. The mean coefficients of variance of the measured radius and the surface area were 2.04 per cent and 3.65 per cent respectively. The results also showed that using a spherical model to calculate the maximal workspace as an index for assessing TM joint impairment is practical.