Through effectively restrain random drift error improve measurement precision of the low-cost microelectro-mechanical system（ MEMS） gyro. The fundamental principle of wavelet threshold de-noising algorithm is researched and the main factors that influence the quality of the wavelet threshold de-noising is analyzed. The types of wavelet function

number of decomposition layers

de-noising threshold and de-noising threshold function are determined. The algorithm of the wavelet threshold de-noising is applied to process actual signals of gyro and the gyro random drift is restrained obviously and good denoising effect is achieved.