Brushless DC(BLDC) motors are widely used for many industrial applications, In view of the problem tha t it is difficult to tune the parameters and get satisfied control characteristics by using normal conventiona l PID controller. a online identification method based on Radial Basi s Function(RBF) has been proposed in this paper. In this method, connection weight of neural network was revised in time according to the speed of motor and phase curr ent, the duty cycle of pulse width modulation (PWM) was adjusted to control the speed of BLDC motor. Conventional PI D and RBF neural network PID algorithm were respectively adopted to make a comparison. the control approach was validated with simulation at first and then was imp lemented with a DSP TMS320F28035. Matlab simulation s and experiment results showed that the proposed approac h has less overshoot, faster response, stronger abi lity of anti-disturbance than the conventional PID controll er.