In this paper, a reduction of forest resources model which taken Zhejiang province as an example was constructed based on the GM model. The model was constructed in order to predict the reduction of forest resources. The results show that the predication posterior variance ratio test was 0.23, the small error probability was 1.00, the GM model were available for prediction of reduction of forest land resources. The models should not be used with reduction of forest land resources caution for general applications. Future studies on reduction of forest land resources prediction models should develop to different area and size so as to predication reduction of forest land resources more accuracy.