Abstract: To optimize the number of automatic guided vehicle (AGV) servicer stations under different cigarette production modes, on the basis of the stochastic service system described by M/G/∞ queuing model, by establishing unit queuing system and taking into account the additivity of Poisson distribution, the formula for calculating the optimal number of servicer stations at different customer source scales was derived. Simulation test and paired comparison indicated that the accuracy of number of AGV servicer stations optimized by M/G/∞ queuing model reached 90%; the model overestimated the actual needs by one service station when customer source scale was small. Adopting the proposed method could estimate the optimal number of servicer stations at different production scales and promote the running efficiency of logistics system for cigarette production effectively.