Considering the issues that the relationship between the faults of screw oil pump existent and fault information is a complicated and nonlinear system, and it is very difficult to found the process model to describe it. The fuzzy neural network has the advantages of both fuzzy theory and neural network. In this paper, a fault diagnosis system with self-repair function for screw oil pump based on fuzzy neural network is presented, moreover, we construct the structure of fuzzy neural network that used for the fault diagnosis of screw oil pump, and adopt the LevenbergMarquart optimizing algorithm to train fuzzy neural network. With the ability of strong self-learning and function approach of fuzzy neural network, the diagnosis system can truly diagnose the fault of screw oil pump by learning the fault information. The real diagnosis results show that this system is feasible and effective.