Abstract: In order to achieve timely and accurately basic information about wetland, which can be applied to the dynamic monitoring and protection of the wetland. The author chose zhalong wetland as the research area, during the process of extracting regional remote sensing information by using the TM image data, DEM data, normalized vegetation index, texture information compound identification index, finally the author classified the types of zhanglong wetland through constructing a decision tree model. For checking the feasible degree of method of classification based on decision tree model, the author made a comparison between the traditional maximum classification of supervision and the decision tree model. The results showed that: the decision tree method based on the index was used to classify, classification accuracy increased by 14.6% , overall Kappa coefficient increased by 0.1751, supervised classification accuracy was improved noticeably. Building decision tree classification which adopted multi-source data for extracting information of inland freshwater mire wetland was a very effective approach.