During a sustained muscle contraction, it is observed that the power spectrum of the myoelectric (ME) signal shifts toward lower frequencies. The mean frequency can be a parameter to monitor the local muscle fatigue. We found that the 2nd order Maximum Entropy Method (MEM) approach provides more accurate and noise-immune mean frequency estimation than the FFT method. The advantage of the 2nd order MEM is simple and fast, and it yields unbiased and consistent estimation of mean frequency of ME signal.