In order to diagnosis the fault of hydraulic system truly and quickly,a fault diagnosis system based on neural network theory is designed for it,the sampled data are real-time processed and the result is displayed visually by utilizing the powerful ability of data processing of the PC,then the diagnosis speed and true-proportion are advanced.When this system is applied to diagnosis the hydraulic system's fault of a vehicle,it gains higher true-proportion and faster speed than the normal fault diagnosis system.