InordertosolvetheformalapplicationofBIMinEPCprojectunderthecon-straints of non-standard contract specifications， to promote the deep integration of BIM andproject implementation， andto realize the expected benefits.This paper establishes theBIM application incentive model based on principal-agent theory.By determining the incen-tive coefficient and BIM application level， the BIM application cost-fee contract design underthe complete and incomplete information is optimized.Based on the first-order derivation ofthe model， the incentive coefficient， the BIM application level， the risk coefficient， the ben-efit-output coefficient and the cost coefficient are discussed.It also analyzes the influence ofoptimal incentive coefficient and optimal BIM application level on both contractors and over-all expected returns.Finally， the effectiveness of the excitation model is verified by the nu-merical analysis of Matlab R2015b throughparameter assignment.The research results show that the BIM application level and incentive coefficient are positively correlated withthe BIM benefit output coefficient， andnegatively correlated with the cost coefficient andrisk coefficient； the overall expected return of incomplete information is greater than the o-ver all expected return under complete information.