Abstract: In order to obtain a larger market share and a better level of profit growth, railway freight companies need to grasp the competitiveness of railway freight rates in the freight market in real time. Comprehensively considering the factors of society, enterprise itself and cargo owner and other factors, based on BP neural network algorithm, this paper studied the method of railway freight rate risk early warning, and established the freight rate risk early warning model. Taking the coal transportation in the bulk cargo transportation of a railway administration group company as an example, the paper selected the relevant data from 2015 to 2017, trained the BP neural network model, and obtained the risk warning results of the railway coal transportation rate. Compared with the actual data, the early warning results have a high degree of fitting, so this method can be used to make a reasonable prediction of the current freight rate risk level, and also play an auxiliary decision-making role for the formulation and adjustment of freight rate policies of relevant railway departments.