The problem of accuracy estimation of the absolute phase recovery in interferometric synthesized aperture radars (InSAR) data processing is considered. A method for accuracy estimation based on the calculation of the standard deviation of the reference phases and the absolute phases measured by the interferometric system in the radar image coordinate system "azimuth - slant range" is developed. The method includes converting the coordinates of the reference topographic height marks from the geographical coordinate system to the radar image coordinate system and then calculating of the transformation parameters between the reference heights and the measured absolute phases using the least squares method. The latter allows one to evaluate the performing effectiveness of individual stages of interferometric processing (multilooking, phase noise suppression, phase unwrapping) and experimentally determine the most effective processing algorithms that provide the best accuracy of the phase recovery and their optimal parameters. The method is tested using the ALOS PALSAR radar images obtained under various imaging conditions.