This paper presents the initial work on implementing a simple control algorithm for a wearable orthotic glove using the myoelectric signals of a stroke patient as the input. The glove is used to assist in the rehabilitation of wrist and elbow. Surface myoelectric signals (MES) are acquired from the biceps brachii and the flexor carpi ulnaris. The MES is processed and temporal features are extracted which are then classified using a multilayer perceptron classfier(MLP) in simulation under real time application constraints. Control signals are generated by the classfier that can be used to actuate servo motors in the glove to facilitate elbow and wrist movement. The simplicity in the glove design and control motivates the patient to use the glove in physiotherapy sessions and at their homes, which is inherent to stroke rehabilitation, for better and faster recovery.