One of the difficulties in surgical navigation based on preoperative image information is intra-operative deformation of the organs consisting of soft tissues. In liver surgery, liver shape is deformed dynamically due to patient respiration, posture change and surgical operations. Such intra-operative liver deformation includes so called large displacement, which requires much computation cost for numerical simulation based on non-linear FEM. Herline [Herline99] reported rigid liver surface registration for such purpose. In this study, we propose a new method of modal representation of liver deformation applied for intra-operative non-rigid registration in image-guided liver surgery. Several experiments with synthetic range data were performed based on error factor analysis.