<p>Method of autocompensation for the effect of constant biases in the output signals of gyroscopes and accelerometers for determining the velocity and coordinates of an object based on the rotation of gyroscopes and accelerometers relative to the object is considered. Two schemes of rotation of the sensitive elements are considered. In the first scheme, all sensitive elements (three gyroscopes and three accelerometers) are installed on a common platform placed in a two-axis gimbal. The second scheme uses two platforms rotating relative to mutually perpendicular axes. Two gyroscopes and two accelerometers are installed on each platform. Mathematical model allowing for the analysis of both rotation schemes is obtained. The analysis of the obtained error models in determining the velocity and coordinates and the simulation performed show the high efficiency of this method for improving the accuracy of measurements. The proposed error model makes it quite easy to evaluate the efficiency of autocompensation. It is shown that the problem may be reduced to integrating the direction cosine matrix that determines the position of the sensitive elements in the reference coordinate system.</p>