Abstract: In this paper the methods and experiments for applying the digital terrain model(DTM) to improve the classification accuracy are introduced.The forest area of Jiu-gong mountains in Hubei province of China was selected as a test area.The author's experimental results indicated that forest types were seriously confused in classified classes when the classification processing had been based on original MSS images only.As soon as the auxiliary data,which refletcs the terrain relief influences to reflecative spectrum of ground feature,were introduced into processing,the classification accuracy could be improved.The average classification accuracy in the test area had been increased to 70.4% in case the combination processing of original MSS images with solar incident angle data was taken.