Abstract: Conventionally, land cover change detection with remote sensing was performed between images matched in Julian calendar dates. The phenology ofplants often brings errors into the final results, and it may make this methodfail in some cases. Detected changes may contain difference in phenology which is not the real change of land cover. To detect accurately crop acreage change, the fluctuation of crop phenology must be excluded. In the paper, the authors deviseda phenological index and applied it in the crop acreage change detection. Matching crop phonological stages allows the users to more strategically choose TM images for the analysis of crop acreage change. This methodology was applied in Shunyi,Beijing as a case study using NOAA AVHRR/NDVI time series data to validate the selection of TM images. The results prove that the method is efficient. In addition, the authors also adopt image difference method to extract the acreage change information of winter wheat from 1999 to 2000, using the theory of the method. The extractionprecision of changing pixel can reach 90%.