Poyang Lake wetland, as an important wetland in the world, is known for its periodic variation of water level in a year affected by the Yangtze River and its tributaries of Ganjiang, Fujiang, Xinjiang, Raojiang and Xiushui rivers. It is very significant to understand the variation of water level for rational exploitation and conservancy of the wetland. This paper, taking Poyang Lake Nature Reserve as a study area, developed a model to simulate the water level and the status of submersion or emersion of wetland using DEM, multi-temporal images and multi-year water level data. First, a new water index FDWI was created to identify the water pixels based on spectrum feature difference between bands 2 and 7 of Landsat TM. The accuracy of classification for water and non-water reaches about 98%. Spatial overlay analysis was performed to obtain the elevations of both water and non-water pixels. Two normal distribution curves were then presented according to elevation statistics. The elevation in the crossed point of two curves was set as the value of water level. With the water level correlation analysis between wetland and Poyang Lake, the regression equations in different lake buffer areas were developed. Furthermore, the model for water level and status of submersion or emersion of wetland was constructed, concerning the topography, buffer areas and water level of Poyang Lake. Finally, the model was validated in estimating the temporal changes of water level of wetland for flood, normal and dry years respectively, as well as in simulating the spatial distribution of both water and non-water pixels of wetland on October 9, 2000. The result shows that the model has satisfactory simulated effect with accuracy above 85%.