Time domain performance of Fuzzy control, including settling time and rise time, is directly relevant to the tuning of variable membership function's slope. In this work we propose an optimal tuning method for variable membership function's slope by simulations. The simulations show that, for fuzzifying the error and the sum of error, increasing the slope of the triangular membership functions can significantly change the initial speed of the controlled BLDC motor drive system. Increasing the slopes for both the left membership function and the center membership function can further improve the starting performance of the controller. But decreasing the slope of right membership function shows worse performance in the fuzzy control of BLDC motor drive system.