Abstract: The traditional methods of forest site type classification is to reflect the relationship of the factors between the forest site system by using a simplified, dependent and indirect mathematical formula, as a result of which, the model classification and evaluation of the effect are not satisfactory. In order to solve the bottleneck of traditional methods, the author employed the LVQ neural network modeling concept, aimed to explore a new set of site classification. The results showed that: LVQ neural network model could better reflect the actual situation of the results of the forest site type classification and provide a reference for forestry operators’ silvicultural planning.