Cogging torque is often a principal source of vibration and acoustic noise in high precision spindle motor applications. In this paper, cogging torque is analytically calculated using energy method to show that Fourier spectra of airgap permeance function and airgap MMF function are the most important design parameters to control cogging torque. To control these functions, stator pole shoe pairing and magnet arc design are proposed to minimize cogging torque. As for optimization technique, genetic algorithm is applied to handle trade-off effects of design parameters. Results show that the proposed method can reduce the cogging torque effectively.