This study aimed to test a possibility of using wireless inertial sensor system in evaluation of gait rehabilitation clinically. Gait movements of a hemiplegic subject were measured with the sensor system during 10 m walking before and after robotic-assisted gait training in a day. Segment inclination angles and joint angles of the lower limbs were calculated from measured angular velocities and acceleration signals based on Kalman filter. Gait event timings were also detected by foot angular velocities and shank acceleration signals. The measured data showed that gait movement changed after the training, which were considered to be improvements by the instruction of the therapist. The robotic-system could be effective for the training based on the instruction. The wireless inertial sensor system would be useful for measurement and evaluation of movements in rehabilitation clinically.