The maneuverability of hydrostatic transmission (HST) is a key factor in improving the performance of agricultural vehicles. In this paper, a new control strategy based on BP (back propagation) neural network is applied to study the dynamic characteristics of the output speed of HST motor. First, the control effects of traditional PID, fuzzy and BP neural network are compared based on the mathematical model of variable pump-quantitative motor. The results indicated that compared to traditional PID control and fuzzy control, BP neural network control can not only effectively suppress the overshoot of the system, but also reduce the fluctuation of motor speed and the adjustment time for system to reach stability. And it has excellent robustness. Therefore, the BP neural network is suggested in this paper to investigate the control effect of the variable pump-variable motor system (VPVM system) with larger motor speed variation range. The results show that this method can be used to achieve stable switching effect and reduce performance loss in segmented control of variable pump and variable motor. For the cases of different equivalent moment of inertia of load, all the motor speed can reach a stable state and the speed fluctuation caused by the load is reduced. The results indicate that BP neural network has a potential advantage in VPVM system control.