Wetland, as one of the weakest ecosystems in the world, is deteriorating rapidly in many regions. Longfeng wetland is the biggest urban wetland in China, located in the inner of Daqing city, Heilongjiang Province. It is divided into two areas by Wolong road, the east is provincial natural reserve and the west is unprotected. This paper describes a remote sensing change detection approach used to assess Longfeng wetland change. Four images were used, i.e., Landsat MSS in August 1979, Landsat TM in July 1990 and in August 2001, LISS-III in July 2006. Linear spectral mixture model (LSMM) method was used to classify the wetland and the results were satisfying. The results showed that the land cover types of Longfeng wetland had been greatly affected by human civilities (such as city extension and fishery), which was more deeply in unprotected area than in natural reserve.