Abstract: As an important data type of digital geometric media, 3D mesh can accurately record the geometric information on the surface of models. The complexity and precision of 3D meshes have been increased, which has brought huge pressure for storage and transmission. In order to improve the compression performance of 3D mesh, a geometry and normal intra-inter prediction encoding method was proposed. Utilizing the correlation of internal components of the normal data (intra-frame) and the correlation between the geometric and normal data (inter-frame), the method was designed to realize the compression of 3D mesh and to effectively maintain geometric characteristics of decoded 3D mesh. Experimental results show that, in comparison with compressing the normal and geometric data respectively, the proposed prediction encoding methods can significantly increase compression efficiency for 3D mesh with normal data.