Abstract: A new idea is presented in this paper which integrates the principal axes method and moment method to normalize and recognize 3D objects. First, criteria for determining the unique principal axes are established which can perform efficiently to normalize the orientations of 3D objects. Then, 3D moment invariants which are position, orientation and scaling independent are derived. Finally, by using the second and third order moment invariants a discriminating vector and criteria for 3D objects recognition are presented. Experimental results of normalization and recognition of twelve 3D objects show the efficiency of our method.