This paper addresses the development of a testing algorithm for pattern-recognition-based strategies to control a myoelectric prothesis. This text describes the structure and components of the proposed algorithm, as well as a process of its testing. The latter included an acquisition of an accompanying EMG, for six different gestures (classes) from seven subjects, as well as its processing, feature extraction, training the classifier and further real-time validation with a robotic hand. The results show that the system provides acceptable classification rates.