Abstract: The microseism signals before and after rock burst on 1300 working face of a coal mine were collected, and sequential characteristics were analyzed. The db5 is used as wavelet basis function to decompose microseism signals into five layers, and the spectral characteristics and energy percentage of each sub-frequency band are obtained. The analysis results show that before rock burst, the number and energy of microseism increase first and then decrease and then increase; precursor signal appeared one hour before rock burst occurred, the energy is mainly distributed in the middle and high frequency region of 62.5-250 Hz, and the energy percentage reaches 55%; when the rock burst occurs, the amplitude increases significantly, with obvious vibration fluctuations, low frequency signals are the main component, and the energy is mainly concentrated in the low frequency range of 0-62.5 Hz, and accounts for about 70% of the total energy. Precursory of rock burst is obtained: the amplitude increases, the frequency of the microseism is significantly reduced, the frequency band develops from high frequency to low frequency, and low frequency signal increases as the time of foreshock occurs closer to the main shock, the percentage of energy occupied by the low frequency signal also increases. Therefore, sharp reduction of main frequency of the microseismic, obvious increase of amplitude and increase of energy percentage of low-frequency signals can be used as the main feature of precursory of rock burst, combined with daily microseismic times and energy trends, rock burst can be predicted.