In an effort to improve the calibration accuracy and ensure the rate sensing performance of gyrowheel, a suitable signal denoising method to extract the useful information from the complicated noise is essential. First, according to the causes of noise, the noise characteristics of gyrowheel were analyzed. Second, the traditional low-pass filter (LPF) was introduced to attenuate the high-frequency periodic noises without influencing the useful low-frequency characteristics. Then the empirical mode decomposition (EMD) method was applied to the filtered signal, and a criterion for selecting relevant intrinsic mode functions (IMF) was presented by using the similarity measurement between probability density functions of IMFs. Combining with the existing thresholding-based denoising technique, a hybrid EMD/LPF denoising strategy was proposed. Simulation results show that the proposed criterion for selecting relevant IMFs was always effective under different SNR conditions. Meanwhile, the proposed method was applied to standard test signals and real signals, and results show the effectiveness and superiority of the hybrid denoising method.