INS/GNSS integration is a well known technique in navigation and, in general, in time-Position-Velocity-Attitude (tPVA) trajectory determination. It is also known, that INS and GNSS are complementary and that the key to their integration is the correct calibration of the Inertial Measurement Unit (IMU) sensors, of the IMU's Inertial Sensor Assembly (ISA) system parameters (axes misalignments and center of navigation), and of the IMU-to-GNSS system parameters (eccentricity vectors). Calibration is performed both at the manufacturer's facilities prior to shipping and “on-the-job,” i.e., together with navigation. Classical INS calibration has two parts: INS error modeling and actual INS calibration once an INS error model is available (also called INS calibration model) either at the manufaturer's or on-the-job. INS error modeling is about understanding the error features — systematic and random — of an IMU and its result uses to be dynamic models — Stochastic Differential Equations (SDE) and white noise amplitude — of the main error sources like biases, scale factors and misalignment angles. INS calibration is about solving for the unknowns (states) of the SDE calibration model together with the unknown states of the SDE of motion. The quality of INS calibration is dependent on the mission dynamics. In classical INS/GNSS navigation, the INS calibration model is dynamics independent; i.e., SDEs formulas and white noise figures are derived following a number of established procedures and then kept fixed. However, it can be shown, that the error features and the INS error calibration model of an IMU change from mission to mission or, more to the point, from one to another type of dynamics. It is therefore natural, to investigate the relationship between IMU errors and motion dynamics and, once this is understood, use this knowledge to improve the performance of INS/GNSS tPVA trajectory determination. In this paper we present a method for IMU error analysis and its first results. It consists of using a reference IMU (REF) together with the IMU under test (IUT) for the simultaneous measurement of REF and ITU signals. First, we show how the sensor random behavior changes with dynamics for tactical and low-cost grade IMUs with respect a navigation grade one. Second, we introduce a procedure to analyze the error of the dynamic dat. The procedure is based on inertial data analysis tools as the Allan Variance applied to the errors of the target IMU with respect to the reference IMU.