Based on analysis of daily maintenance cost data of Yuhe Highway,Yuqian Highway and Qiwan Highway and by using regression analysis,the relevance model between maintenance cost and using duration of Chongqing highway was constructed,the optimal opportunity for Chongqing highways to implement preventive maintenance was obtained,and BP neural network was used to construct predictive models in order to attain the goal for forecasting the cost needed in Chongqing highway maintenance in the future.