The purpose of this research is to improve efficiency of rehabilitation exercise programs with smart-bar for patient who have musculoskeletal disorders such as turtle neck syndrome, and scoliosis. The smart-bar consists of two parts which is mainly composed of six-axis, acceleration sensor, to measure a movement of arms, hands, shoulder and head. The smart phone is shown the program of daily exercise on screen and is collected the data of motion movement from both smart-bar and camera of smart phone. The image from camera of smart phone was analyzed a movement of arms, hands, shoulder, and head, with various algorithm such as block matching method, horn-schunck algorithm, and lucas-kannade algorithm. To test the rehabilitation exercise system, 6 males who have any physical disease on musculoskeletal system for two years were enrolled in the study. The exercise program consists of three basic motions which are seated side stretch (right, and left), and rhomboid stretching. When volunteers exercise without the real time motion analysis system, the accuracy of exercise was 82.25±1.23%, 82.42%±0.93%, and 80.48%±3.25%, for right, left, and rhomboid stretching, respectively. When volunteers exercise with the real time motion analysis system, the accuracy of exercise was 93.37±1.15%, 93.82%±0.58%, and 95.64%±1.36%, for right, left, and rhomboid stretching, respectively. When they exercise with real time motion analysis system, they experienced more feeling of satisfaction.