The range migration problem, a key problem in SAR (synthetic aperture radar) signal processing, is addressed. The long distance from the sensor to the target makes the problem so serious that one cannot get a fine image without correction. The authors have studied the generation of the range migration and have derived a new method to correct it. Because the correction is done during the crosstrack direction pulse compression instead of after it, the process is simpler and more reliable than previous methods. If a high-resolution image is required, one can compensate the range curvature in the frequency domain. The proposed method was simulated on a VAX workstation and achieved satisfactory results in the processing of real space SAR data.<>