Various types of indices commonly used to illuminate the properties of evaluated unit just reflected its situation from a few of the sides; therefore, simple using of these indices to proceed mathematical analyses and to do regionalization research would usually deviate from the right direction. Taking the basic principles for regionalization as a directing framework, the use of mathematical method would be more reasonable, apart from getting more typical and exacter data. The use of classical principles for regionalization in mathematical analyses is helpful to the selection of evaluated unit, as well as explanation of the result of mathematical analyses. Cluster analysis based on factor analysis could carry out some principles for regionalization, such as the linking of integrated analysis with dominant factor, comparative consistency, and conjugate. The rational use of mentioned method would improve the regionalization study. By using the mentioned principles and multivariate statistical analysis, selecting 30 provinces, autonomous regions and municipalities as evaluated units, China could be divided into 5 regions and 12 sub-regions concerning its land use degree and benefit.