Abstract: To solve problems of multimodal traffic assignment on a composite traffic network,the network is trans-formed into a multi-level structure,and a model of traffic assignment together with its constraints and solving algorithms are proposed.Firstly,the multimodal composite network is explained by a multi-level network structure.A three-level network is taken as a case study,the first level is for mode choices in consideration of combined travels;the second level is a state-augmented multi-modal network based on travel logic;the third level is a road network for cars.Secondly,path impedance of a multi-level network is analyzed including link impedance and transfer impedance.An N-L model for traffic assignment on this network is developed.According to the actual structure of the composite network,the efficiency of solving this model is improved through adding constraints of effective paths,effective transfers,and typical combined travels.Thirdly,the Method of Successive Averages(MSA)is used to solve the problems of the three-level network.Fi-nally,a numerical example is proposed based on traffic assignment on a composite network with three types of traveling modes.The model iterates seven times to meet the convergence conditions.The results show that in the case study,rail transfer has certain competitiveness,accounting for 40% of the total travels.The reliability of results is verified through SP investigation.This method takes path choice,transfer node choice,and composite travel choice into consideration and therefore overcomes drawbacks of the traditional four-stage method in multimodal traffic assignment.