Direct-driven spindles have no mechanical transmission trains and gears, and are the key actuators for computerized numerical control machine tools. These magnetically suspended rotor systems are required to provide fast response and high precision. However, these systems are non-linear and strongly coupled. The traditional proportional, integral, derivative (PID) control method has been widely used for such systems owing to its relative simple realization. However, the tracking, disturbance rejection and robustness properties of the controlled plant may not be satisfied. To solve these problems, this paper presents a decoupling control strategy based on an inverse system scheme and combines it with the internal model control method to guarantee system robustness to the parameter uncertainty and external disturbance. By introducing an inversion of the magnetically suspended rotor system into the original system, a new pseudolinear system is developed. It can be shown that this addition effectively eliminates the influence of the unmodelled dynamics, and improve the accuracy and robustness of the whole system. The simulation and experimental results show that when compared with the traditional control scheme, the proposed control scheme provides good decoupling and robustness performance for the magnetically suspended rotor system in different operating conditions.