Mineral liberation assessment of ore particles generally involves two-dimensional (2D) measurement of particle cross-sections, and this 2D measurement inevitably results in error due to stereological bias. This bias is considered to be strongly influenced by the internal mineral structure of the particles, but its overall characterization, though investigated conceptually, has hitherto lacked a quantitative foundation. In this study, a systematic numerical study was conducted on particles with an internal mineral structure based on Voronoi modeling, in which three-dimensional degrees of liberation were compared with their 2D counterparts, for random particle cross-sections. In this way, for the first time, overall quantitative characterization of stereo logical bias, in terms of the volume fraction and size of the mineral phase, was achieved.