One of the important challenges in design of the PM electrical machines is to reduce the cogging torque. In this paper, in order to reduce the cogging torque, a new method for designing of the motor magnets is introduced to optimize of a six pole BLDC motor by using design of experiment (DOE) method. In this method the machine magnets consist of several identical segments which are shifted to a definite angle from each other. Design of experiment (DOE) methodology is used for a screening of the design space and for the generation of approximation models using response surface techniques. In this paper, optimization is often solved via surrogate models, that is, through the construction of response surface models (RSM) like polynomial regression. The experiments were performed based on the response surface methodology (RSM), as a statistical design of experiment approach, in order to investigate the effect of parameters on the response variations. In this investigation, the optimal shifting angles (factors) were identified to minimize the cogging torque. A significant reduction of cogging torque can be achieved with this approach after only a few evaluations of the coupled FE model. (C) 2011 Elsevier Ltd. All rights reserved.