Abstract: Based on analyzing the existingmethods for evaluation of sustainable development of coal enterprises, an evaluation-index system was established. A neuralnetworkmodelwithmulti-hierarchic structurewas built based on error back-propagation (BP) algorithm. The conjunction weights of the neural network are modified layer by layer from output layer to input layer in the process of neural network training to reduce the errors between the anticipated and actual outputs. A case study was conducted and itverified the validity and accuracy of the proposedmethod.