Surface Electromyography (sEMG) has been widely applied in different fields, such as human-computer interaction and medical rehabilitation. This paper deeply studies the theoretical research of muscle fatigue analysis and scientific experiments of sEMG, especially related to muscle fatigue. It provides a theoretical basis for not only building an assessment method of muscle fatigue but also testing the relationship between the value of wavelet packet entropy and the complexity of signal frequency. Additionally, the field value of muscle fatigue has been uncovered. In this research, the experiment tests the energy variation of wavelet packet entropy when testing the muscle contraction by the sEMG signals of brachioradialis. It means that the experiment decomposes frequency band, and then sEMG is analyzed by the entropy and distribution of energy. The result of experiment indicated that the index of wavelet packet entropy has an efficient and quick performance on analyzing complexity of signal system. In the experiment of muscle fatigue, wavelet packet entropy can present high accuracy, instant reaction, stronger consistency and reliability, which is significant for the achievement of the real-time monitoring and clinical research of bioelectrical signals.