In this paper, we propose an algorithm to calculate pose normalization for 3D mesh models based on an analysis about rotation transformation in the theory of analytic geometry. The method first evaluates an angle and an axis for rotation transformation of each triangle in model. Then, optimal rotating axis and angle are calculated by using probability theory for the whole model and a global rotation matrix is consequently constructed. Finally global rotation transformation defined by the matrix can be obtained and used to transform the model to normalized pose. The experimental results show that our algorithm can calculate pose estimations of models efficiently and automatically.