Due to the complex calculation process of traditional methods in the entire process cost prediction of construction projects, it doe not consider the complex interaction relationship between multiple influencing factors during the cost formation process, so a construction project cost prediction model based on Bayesian networks is proposed. This method integrates project experience, historical information, and actual cost performance data sequences observed during project implementation, to determine the degree of influence and interrelationship of each influencing factor on the cost components. Then, an explanatory structural model and Bayesian network are used to establish a cost prediction model for construction projects, and the entire process cost of the construction project are calculated and output through analysis functions. Finally, the application effect of the model in an example is explored, and the results show that the various costs and total costs acquired by Bayesian network conditional probabilities are basically consistent with the budgeted costs of original plan. The model not only has feasibility and high accuracy, but also has a positive impact on the probability and degree of low project cost risk events.