A stroke ischemic is one of a stroke type that commonly occurs in people with a stroke. This is usually followed by symptoms which are a weakness or paralyze of a limb. The purpose of this study is to practice the stroke patient independently using a robotic exoskeleton based on electromyography (EMG) signal. This study was an experimental research with a pre and post design. The results of conventional exercise were used independently in addition to the robotic exoskeleton method during two weeks. Samples were taken based on the patient who had MT-3 (Manual Scale Testing) measurement scale. The EMG signal was collected from biceps during the exercise. The EMG signal was extracted using root mean square (RMS) feature which represented the energy of the muscle. The EMG data were analyzed using single factor ANOVA to obtain a significant difference of RMS. Results indicated that there was no significant difference of RMS with a p-value of 0.85 (alpha= 0.05). The used of the robotic exoskeleton method for rehabilitation of stroke patient was not effective for a short time assessment but more preferred as an observation of the EMG activities during the exercise for the post-stroke patient.