In this paper, steering control based on the model predictive control (MPC) framework was investigated for vehicle automatic lane keeping  system (ALKS). First, vehicle lateral dynamics and tire cornering properties were analyzed. The lateral distance error, yaw angle error and the derivates were chosen as the state vector, the steering angle for the front wheel was considered as the control input, and the state space model was developed. Based on the model, the performance index and system constraints for ALKS were established, the smooth desired reference trajectories were introduced, and the MPC control strategy was designed for steering control. Simulation experiments show that the control algorithm eliminates the lateral distance error and yaw angle error quickly, guarantees that the vehicle moves along the center of the lane, smoothes the dynamical responses, and performs well in adaptability and robustness at different speeds.