The inertial navigation system has a requirement on HRG that it should have a high accuracy in the wide dynamic range,while the HRG scale factor obtained by the conventional identification method based on the least square fitting method is not very accurate thus can result in a big navigation error.To improve the accuracy of HRG scale factor,an identification method based on Neural Network is proposed in this paper.By using the gradient descending momentum learning method of neural network error back propagation,the HRG scale factor can be calibrated.The experiment has verified that this new identification method is feasible. It provides the basis for improving the HRG accuracy and descending the navigation error of inertial navigation system.