During the course of rehabilitation hemiplegic patients who have Chedoke McMaster Stages of Motor Recovery scores 4 and 5 measured three weeks after onset of stroke often improve their arm and hand function to the point that they can later use it in the activities of daily living (ADL) (1). These patients can be considered to have mild arm and hand paralysis since they can grasp objects and manipulate them with minor restrictions in the range of movement and force. On the other hand, hemiplegic patients who have Chedoke McMaster Stages of Motor Recovery scores 1 and 2 measured three weeks after onset of stroke, during the course of rehabilitation seldom improve their arm and hand function, and when they do, the improvements are not sufficient to allow these patients to use the arm and hand in ADL (1). These patients can be also described as patients who have severe arm and hand paralysis. Patients with severe arm and hand paralysis cannot move their arm and hand voluntarily at all or have very limited voluntary movements that cannot be used to carry out ADL. In recent years a variety of treatments such as constraint induced therapy, functional electrical therapy, biofeedback therapy, and robotics assisted therapies, were proposed which main objective is to improve reaching and grasping functions in subjects with unilateral arm paralysis. These therapies have shown encouraging results in patients with mild arm and hand paralysis. However, the efficacy of these therapies was limited when they were applied to patients with severe arm and hand paralysis. This article describes a new rehabilitation technique that can improve both reaching and grasping functions in hemiplegic patients with severe unilateral arm paralysis. A neuroprosthesis that applies surface electrical stimulation technology was used to retrain hemiplegic patients who had severe arm and hand paralysis to reach and grasp. The neuroprosthesis was applied both to acute and long-term hemiplegic patients. Patients who were treated with the neuroprosthesis were compared to those patients who were administered only standard physiotherapy and occupational therapy appropriate for hemiplegic patients with unilateral upper extremity paralysis (controls). The treated and control patients had approximately the same time allocated for arm and hand therapy. After the treatment program was completed, the patients treated with the neuroprosthesis significantly improved their reaching and grasping functions and were able to use them in ADL. However, the majority of the control patients did not improve their arm and hand functions significantly and were not able to use them in ADL.