Electromyography (EMG) based control is growing to be a part and parcel of the assistive devices technique today. Nonetheless, high sensitivity of EMG to the muscular fatigue affect the performance of EMG signal as a control input to the assistive devices. In this paper, the influences of muscular fatigue on the ankle joint motion recognition have been investigated. Three shank muscles and two ankle joint movements have been involved in this experiment. In order to assess the muscular fatigue on the motion recognition based on EMG signal, Multilayer Percepteron (MLP) and K Nearest Neighborhood (kNN) were employed. The outcomes of this experiment showed the drastic change in the recognition accuracy of the two ankle joint movements before and after inducing muscular fatigue. The overall classification accuracies for all subjects before fatigue were 97.7% and 97.4% for MLP and kNN respectively. Whereas, the overall classification accuracies for all subjects after fatigue were 92.89% and 91.09% for MLP and kNN respectively.