Abstract Electromyography signals are commonly used to control prosthetic or orthotic devices. The electrodes attached to the skin capture the muscular activity coming from neuromotor signals. Robotic prostheses help to improve the quality of life of people who suffer from physical disabilities such as amputations of lower or upper limbs. However, these prostheses are quite expensive, and the operation of the robotic prosthesis depends on the contraction muscles that generally uses two movements, contraction and relaxation (closing and opening). In this paper, we propose the characterization of hand movements by means of an inexpensive electromyographic sensor, the MyoWare AT-04-001, which acquired neuromotor signals due to muscle contraction in the forearm. Four movements were recorded: flexion, extension, opening and closing of the hand. Hand movements were performed at one movement per second, as indicated by a metronome. Signals were acquired with an Arduino one at a sampling frequency of 200 Hz for 30 s. Using time-domain, frequency-domain and nonlinear measures we characterized the electromyography signal for each movement. We observed that the closing and extension of the hand produced the greatest amount of variance and entropy of the electromyography signal as well as the greatest energy in the frequency domain. These results were related to the location of the electrode in the forearm, since the sensor was placed in the muscles involved in the execution of these movements.