Abstract: To solve the problem of traditional fault diagnosis for tea garden tractor with serious delay during collecting fault information, a remote fault diagnosis system（RFDS） of tea garden tractor was proposed based on fuzzy neural network（FNN）. Combining the advantages of fuzzy algorithm and neural network, through the real-time monitoring of vehicle operation information, the vehicle information was remotely processed to obtain vehicle potential fault and real-time fault information for avoiding a large number of detections in traditional methods, and more time was gained for maintenance and improving production efficiency. Based on the simulation software of Carsim, the vehicle operation was simulated in real time, and the algorithm model was trained and verified by training set and test set. The results show that FNN diagnosis algorithm can meet the performance requirements of the system, and the accuracy can reach more than 90%. The tractor fault in tea garden can be diagnosed remotely. RFDS technology also plays an important role in the development of vehicles. The performance of vehicles can be evaluated through remote fault diagnosis system in the R ＆ D stage of vehicles, which saves a lot of human resources.