The absence of the human hand causes difficulties for the patient to perform simple daily tasks. However, anthropomorphic hand prostheses can help users regain object manipulation and grab functions. For underactuated prostheses, which are generally lighter and have a lower cost of production, precision in controlling movements is a crucial factor. In this context, this paper proposes and computationally investigates the control strategy with a complete command vector for the flexion and extension movement of the index finger of an underactuated hand prosthesis. Compared to conventional PI and PID controllers, this control topology allowed for better system responses, producing considerably lower position errors than the classical controllers.