The Micro Electro Mechanical System (MEMS) gyroscope have a wide range of applications because of the virtue of compact size, low cost and power consumption, and mass production. Since the MEMS gyroscope usually has the large drift, how to effectively reduce the random drift error is crucial in improving the gyroscope precision and performance. In this paper, we propose a new de-drift method based on the recently developed empirical mode decomposition (EMD). The proposed EMD-based method is able to reduce the noise term of random drift. The method is also compared to a related approach based on the Wavelet filtering method. Experiment results indicated that the drift noise parameters were reduced and noise signals were adequately restrained.