The prices of main materials for highway engineering are complicated variables which are affected by many factors. Application of improved BP algorithms in price prediction of highway engineering main materials was discussed. The specific methods are as follows£o First, the shortcomings of traditional BP algorithm were pointed out and then its improved algorithms were put forward. Then, the main influencing factors of highway engineering main materials were screened out with correlation analysis method. On the basis of determining the structure of BP neural network and selecting training functions, a price forecasting model based on improved BP algorithms of highway engineering main materials was established. Finally, combining with the case of forecasting the prices of stone chips in Hefei, traditional and three improved BP algorithms were used and the forecasting results were analyzed. The results indicate that forecasting by applying improved BP neural network is feasible and effective. The forecasting errors can be reduced to less than 6% and training steps can be reduced by 95% percent using self-adaptive study speed algorithm and additional momentum algorithm simultaneity.