An iterative learning servo controller with RBF network feed-forward position compensation is proposed to control the mover of permanent magnet linear servo system to track reference position signal. The structure of RBF feed-forward position controller is analyzed in detail. The effective estimation of nonlinear load is realized by higher-order D-type iterative learning identification method. The feedback position control law was made up of load identification and feedback error signal. Finally, the simulation results demonstrate that this method can assure system obtains good servo performance.