Abstract: Forest growth under management is a gradual process, which is resulted from interaction of nature and man disturbance. Different forest succession stages have specific attributes, playing a guiding role in forest management. Appropriate management can positively affect forest growth, otherwise, the effect is little or even negative. Consequently, identifying forest succession stage has significant influence on improving forest management level. Five forest stages, i. e. forest establishment stage, competition differentiation stage, selection stage, close-to-nature sand naturalness permanent forest stage, were presented based on previous researches. Stepwise discriminant (Wilks' Lambda) was used to select important indices among initial 13 stand indices. Age, average DBH of dominate tree species, ratio of tree volume to total volume, stand basal area, and DBH variable coefficient were selected to classify forest stage. Fisher discriminant function and Bayes discriminant function were established referring to the five indices mentioned above, the classification results were 94% of training samples correctly classified, 89% of verification samples correctly classified, and ratio of tree volume to total volume is the best index to identify forest stage.