The position tracking and control method using the bioelectrical signals attracts attentions as expectable technology. The purpose of this research is to construct a practical master slave system, which uses electricmyograrm (EMG) signals. Signal processings of EMG signals are performed using a linear multiple regression model, which can learn parameters in a short time. Using this model, joint angles are predicted, and the motion pattern discrimination is conducted. Several experiments were conducted to verify the validity of this technique. First, the motion pattern discrimination from EMG signals was conducted. Discriminated motions were flexion, extension, pronation, and supination at an elbow joint Second, the experiment using a robot arm was conducted In this experiment, an elbow joint angle was predicted from EMG signals. From these experiments, the usefulness of processing EMG signals with a linear multiple regression model was proved.