To effectively restore the fine motor functions of the forearm and hand of stroke survivors and patients with traumatic brain injury (TBI), recent studies have proposed an active rehabilitation concept based on the pattern recognition of electromyography (EMG) signals to decode the motor intent of the patients. The results from these studies suggested that pattern recognition of EMG signals associated with the limb motions could potentially aid the development of active rehabilitation robots. To obtain richer set of neural information from multiple-channel EMG recordings, this study proposed a muscle synergies based method for motor intent identification from high-density CP EMG signals recorded from eight TBI subjects. For baseline comparison, the linear discriminant analysis (LDA) based pattern recognition approach was also examined. The outcomes show that the proposed muscle synergy based method outperformed the commonly used LDA with more centralized distribution of motion classification accuracy across all the TBI subjects. And such an increment in accuracy suggests the feasibility CP of using muscle synergies for neural control in active rehabilitation for TBI patients. © 2017 IEEE.