Abstract: The traditional methods for estimating pixel offsets of two SAR images use the orbit offsets of several points to fit the orbit offsets of the whole image by using the least squares rule,and then removes them from the total offsets.However,the procedures of data processing are quite complicated.What's more,without GCPS,this method contained many kinds of error sources which lead to low precision.To overcome these limits,this paper presents a new simple data processing chain for calculating co-seismic offsets which need no GPS data.In order to compare the precision of two methods,we take the Ms6.6 Bam earthquake occurred in 2003 as an example.First,we processed two SAR images acquired before earthquake with the traditional and the new pixel offset methods respectively.The results show an improvement of 0.224 pixels in precision for the new method.Second,we applied the new and the traditional method to process two images acquired before and after the earthquake.The results indicate that the new method has a higher efficiency and precision.