Model predictive control algorithms have recently gained more importance in the field of wind power generators. One of the important categories of model predictive control methods is improved deadbeat control in which the reverse model of generator is used to calculate the appropriate inputs for the next iteration of controlling process. In this paper, a new improved deadbeat algorithm is proposed to control the stator currents of an outer-rotor five-phase BLDC generator. Extended Kalman filter is used in the estimation step of proposed method, and generator equations are used to calculate the appropriate voltages for the next modulation period. Two aspects of proposed controlling method are evaluated including its sensitivity to generator parameter variations and its speed in following the reference values of required torque during transient states. Wind power generators are kept in mind, and proposed controlling method is both simulated and experimentally evaluated on an outer-rotor five-phase BLDC generator. (C) 2014 Elsevier B.V. All rights reserved.