Abstract: Main factors and indicators that affect water inflow of a coal mine of Shaanxi were analyzed according to its hydrogeological conditions. BP neural network and large-well method were used to predict water inflow of different working faces of the second and third districts, and results of the two methods were compared. The analysis results show that the predict result of BP neural network model is more accurate, it can be used as basis for development of mine drainage and pressure decreasing program.