Abstract: The forest communities with rich species composition in Daqinggou was a special forest ecosystem type. Communities were classified by Sum of Squares of Deviations of systematic cluster method with importance values of trees, shrubs and herbs in 38 plots. The results by this method and Fuzzy Clustering were compared. The forest communities were divided into four associations, and character of every community was explained. The functions of every species in every community layer were comprehensively considered. So the classification result was meticulous and objective, it could be regarded as the fundamental materials for Protective Area Office to scientifically protect and manage forest resources. The results also show that methods of the Sum of Squares of Deviations and Fuzzy Clustering are suitable to community classification in Daqinggou.