BP (Back Propagation) algorithm, the full name of back propagation algorithm, is the most commonly used training algorithm in neural networks. This paper aims to discuss the construction and application of construction project cost (CPC) prediction model based on BP algorithm. By collecting and analyzing a large amount of construction engineering data, this paper proposes a construction engineering cost prediction model and verifies it experimentally. This paper uses the method of experimental comparison for analysis. Experimental results show that the material quantity estimation model based on BP algorithm has good performance. On the test dataset, the estimation results for the three different examples all perform well. Among them, the material estimation error under the BP algorithm is about 10% (5–14%), the maximum is 13.6%, and the minimum is 5.5%. This model can effectively solve the problem of CPC forecasting.