This paper presents a fractional model identification for a billet unloading robotic arm’s positioning system. First, an integer order model is obtained using a graphical identification method based on a set of experimental data. The experimental data represents the robotic arm’s position, measured using an encoder, at a constant billet displacement. The integer order model was obtained based on the overall performances of the measured robotic arm’s step response. The mean square error between the measured data and the model step response is high, thus, in order to decrease the error and to obtain a more accurate model, a fractional order model is determined using an iterative procedure.