This paper presents a cascaded nonlinear predictive control of a permanent magnet synchronous motor (PMSM) drive. The Taylor series expansion is used to carry out the prediction defined on a finite horizon. However, it's well-known that this control strategy cannot remove completely the steady state error under mismatched parameters and load torque. Then, the full knowledge of machine parameters and operating conditions must be used to compute the control law. For that reason, a disturbance observer is designed for the estimation of the offset caused by the parameters uncertainties and external disturbance. By taking into account the offset observer in the global control system design, an equivalent cascaded proportional integration (PI) action is obtained. The performances of the proposed control strategy are analysed by simulation. The obtained results show the effectiveness of the proposed control strategy regarding trajectory tracking and disturbance rejection.