Muscle fatigue and early tiredness are among the common findings of. many conditions that affect neural and muscular systems. Quantitative measurement of muscle fatigue and its possible result, tiredness may provide an objective criterion for the follow-up and observation of the patient's treatment stages. In this work, a method is presented to evaluate the muscle fatigue in human arms by using surface electromyography (EMG) signals. EMG signals are recorded from two muscles during isometric contraction under constant force values, from eleven healthy subjects. Then, features that are capable of characterizing the EMG signals are extracted. Alternative to the previous methods in the literature using median frequency, behavior of EMG signals is characterized by higher order statistical moments. Preliminary results show that this approach may be a valuable contribution to EMG signal-muscle fatigue research.