In order to improve the precision of path tracking, a self-tuning model control method for farm machine navigation was presented. The thought of the two-level control was induced to design the navigation controller. As the first level controller, the nonlinear kinematics model and feedback linearization theory were used to design the control law, and counting formula of the system controlling quantity was given. In order to solve the imprecision problem of the nonlinear kinematics and improve the stability of the navigation system, a second level controller was needed. As the second level controller, the fuzzy controller used the fuzzy control method to adjust system control quantity of model controller online and adaptively. The results of the path tracking experiment for the agricultural machine showed that the control method kept the advantages of model controller for tracking the straight and solved the problems of model controller in a curved path tracking. When it ran at the speed of 1.0m/s, the maximum lateral deviation of straight path tracking was less than 0.0649m, the maximum lateral deviation of curved path traced was less than 0.1857m.