The motion error generated by airborne SAR platform during flight will cause the image quality to degrade. Extracting the motion error using measurement data is one of the important means of airborne SAR motion compensation. Through the derivation of airborne SAR motion error theory, a method of airborne SAR imaging motion compensation based on moving least square is proposed. This method processes the north, east and up 3D velocity information recorded by the inertial navigation system, and then processes it. The three-dimensional spatial information of the radar platform is obtained, and then combined with the backward projection algorithm to complete the SAR imaging motion compensation. This method is to improve the least square method, so that it does not need to piecewise estimate and smooth the data, and can also ensure the accuracy of the fitted data. Experimental results verify the accuracy and effectiveness of the method for estimating motion errors.