Path following is a fundamental problem for unmanned air vehicles (UAVs). In this paper, a Lyapunov stable path guidance law is designed to generate the course control command for a small UAV. The control parameters of the guidance law is optimized by the nonlinear model predictive control (NMPC) techinque. The relative simulations with the three different approaches: Lyapunov stable controller with parameters optimizated (LC_NMPC), the normal NMPC and the Lyapunov stable controller with constant parameters (LC) are compared with each other. The simulation results show that the LC_NMPC can reach good path following performance with shorter rise time and zero steady error.