The thesis describes testing of various calibration procedures of the magnetic inertial measurement system (MIMS), with which certain deficiencies, and errors can be eliminated, and consequently measurement can be improved. Errors of individual sensors are in most of the cases unknown, therefore, I first determined the characteristics of random processes (white noise, drift etc.) using the Allan variance method. Every sensor has its own errors that are modelled in the process of calibration with an appropriate numerical model. Therefore, the accuracy and reliability of calibration procedures is different. The performance of six different calibration procedures in known conditions was empirically evaluated using a commercially available MIMS (SU6DOF). A series of tests for accelerometer and gyroscopes have shown that the best method in terms of accuracy, simplicity, and errors was presented by Stančin. A comparison of various calibration methods for a magnetometer however, showed no difference in total average error. Only a difference in the size of the error among the static and dynamic calibration of magnetometer was distinct.