One of the main problems in developing active prosthesis is how to control them in a natural way. In order to increase the effectiveness of hand prostheses there is a need in better exploiting electromyography (EMG) signals. After an analysis of the movements necessary for grasping, we individuated five movements for the wrist-hand mobility. Then we designed the basic electronics and software for the acquisition and the analysis of the EMG signals. We built a small size electronic device capable of registering them that can be integrated into a hand prosthesis. Among all the numerous muscles that move the fingers, we have chosen the ones in the forearm and positioned only two electrodes. To recognize the operation, we developed a classification system, using a novel integration of Artificial Neural Networks (ANN) and wavelet features.