The supply chain risk forewarning index system was established by analyzing the influencing factors of supply chain risk,and the error of the system was decreased to the required accuracy by correcting the weights of model with BP neural network self-learning repeatedly.According to the output of the model,the key risk factors which produced supply chain risk could be found out by contrasting the weight of each layer.28 typical supply chains in Hebei Province were taken as example,25 of which were used to train the risk early-warning system,and the others to test.The result shows that the supply chain risk forewarning index system can achieve 90% above of prediction accuracy and find more practical key risk factors through the network weights analysis.