Based on the analysis of the static error model for RLGSINS, a relaunch calibration approach of RLGSINS is investigated from the situation of the real system application. The error coefficients selfcalibration, which is independent of the external baseline information, has been completed by making the most of the selfmeasurement data of RLGSINS. The experimental results show that the proposed error coefficients selfcalibration approach of RLGSINS can meet the prelaunch selfcalibration precision demand and is easy to perform, and has comparatively strong practicability in engineering.