Abstract: The common methods of extracting information of forest using ETM+remote sensing image were discussed in this paper. According to the analysis of eigenvalue and relativity, it was found that bands of TM3, TM4 and TM5 had small relativity and contain ample information. Bands of TM3, TM4 and TM5 were merged with PAN bands using Brovey transform. NDVIR was obtained by computing Normalized Difference Vegetation Index for bands of TM4R and TM3R of merged remote sensing image. Then the combination image of fusion remote sensing images and NDVIR image was classified by method of maximum likelihood. The forest cover rate was computed to be 38.63% according to the classification.