Abstract: In order to accurately assess the risk of water gushing when a tunnel crossing through faults and avoid the occurrence of surge disasters, water gushing risk assessment is carried out based on artificial neural network(ANN).On the basis of extensive reference to various norms and documents, a risk evaluation index system was constructed based on 13 main risk factors affecting the occurrence of surge disasters in fault zones.The ANN was trained and tested using engineering example data and the backpropagation algorithm.When the network parameters were optimized in several trainings, the risk assessment on water inrush ina tunnel fault zone was realized.The generalization test results show that the ANN can predict water gushing risk level quickly and accurately.The proposed method overcomes some problems of traditional risk assessment methods, such as subjectivity, uncertainty and complex calculation, and provides a reference for risk assessment of water gushing from fault zones in tunnels.