Aiming at the problem that the nonlinearity of MEMS gyroscope at low speed range is very strong, the conventional error calibration method cannot meet the practical application, a calibration and compensation method based on the back propagation (BP) neural network is proposed in this paper. A set of rate points based on the number of priority is designed and 12 groups of rate experiments are carried out with the threeaxis turntable. The least square method is used to calculate the elements of the traditional mathematical model for calibration. We use the outputs of three MEMS gyroscopes and the actual rate of the threeaxis turntable as the sample to train the BP neural network to build the model. Also, we compare the compensation effect of two methods. The results show that both the traditional method and the BP neural network can compensate the output of the MEMS gyroscope effectively. However, the compensation effect of the neural network is about 3 times higher than that of the traditional method at the low speed range.