The team of the Department of Laser Measuring and Navigation Systems of SPbGETU LETI proposed a method for diagnosing surface defects based on inertial measurements. This approach was later introduced into the AVIKON-03M flaw detector car produced by the JSC “Radioavionika”, which was equipped with a set of micromechanical inertial modules for this purpose. The system operation confirmed the possibility of measuring the surface defects' parameters. However, during the analysis of the inertial data it was very problematic to distinguish the signals received when moving along a rail defect from those received while passing a rail joint. That raised the question of identifying and classifying the detected defects. For this purpose, it is now proposed to perform data fusion of measurements obtained from inertial sensors and a magnetometric subsystem installed on the car. This data fusion is the subject of this work.