In an autonomous vehicle, the reference lane is continually detected by a machine vision system. Then the vehicle is steered to follow the reference yaw rates which are generated by the deviations of lateral distance and the yaw angle between the vehicle and the reference lane. To cope with the steering delay and the side-slip of a vehicle, a PI controller is introduced for the yaw rate feedback. It is tuned by the simulation that the vehicle is modeled as 2 DOF and 79 DOF and verified by the results of the actual vehicle test. The lateral control algorithm by the yaw rate feedback has good performance of lane tracking and passenger comfort.