Abstract: In order to select optimal paths quickly and efficiently, a multi-objective multimodal multi-commodity routing model with time windows was proposed on the basis of the existing model. A two-layer search algorithm was designed to solve the model. In the first layer, a revised Martins label setting algorithm and two time constraints are combined to calculate the effective paths based on the labels of paths. In the second layer, the effective solution of the above algorithm is considered as the initial solution, and optimal paths are obtained by removing the paths which do not meet the three restrictive conditions in transport time of cargoes, transfer times, and capacity limitation of transport mode. Then, a technique for order preference by similarity to an ideal solution (TOPSIS) was used to calculate integrated weights by combining different cost weights and time weights together, so as to find the optimal paths with the maximum integrated weight value. The result of an application example shows that, compared with the existing labeling algorithm, the proposed algorithm can reduce the search space and avoid generating invalid solutions by including capacity limitation of transport mode; compared with the Lagrangean relaxation method which can only generate the lower and upper bounds on the optimal solution, the proposed algorithm can obtain exact solutions within 30 min, computing time being reduced by about 75%.