Aiming at the uncertainties existing in the permanent magnet linear servo system, an intelligent second-order sliding mode control (ISOSMC) method is proposed to achieve high precision and strong robustness. First, the dynamic model of PMLSM with uncertainties is established. To conquer the uncertainties, a second-order sliding mode control (SOSMC) method based on the PID sliding surface is designed. Additionally, to further reduce chattering caused by sign function, a sinusoidal saturation function is applied in the control law of SOSMC. Due to the value of the uncertainties is di cult to obtain, a recurrent radial basis function neural network (RBFNN) uncertainty estimator is introduced to estimate the uncertainties and a robust compensator (RC) is developed to reduce the approximation error of recurrent RBFNN. To guarantee the performance of the ISOSMC system, Lyapunov function is employed to prove the stability. Experimental results demonstrate ISOSMC possesses favorable tracking performance and strong robustness against the parameter variations and load disturbances.