Cooperation in logistics distribution among enterprises can significantly reduce the cost and emission of logistics delivery. Key research questions are how to model cooperation in logistics distribution and develop a cost sharing method. Traditional cost sharing methods need to calculate the cooperative cost for all sub-coalitions，which is equivalent to solving 2N - 1 complex vehicle routing problems for calculating the cooperative cost of 2N - 1 sub-coalitions ( N is the number of enterprises) . This paper analyzes the properties of the cost sharing problem in collaborative distribution by modeling a multi-cooperation vehicle routing. Based on the classical Shapley cost sharing method，a Binary Tree ( B-T) Shapley method is proposed. The new method can not only reduce the computational complexity of cost sharing from O ( N2 2N ) to O ( N2 log2 N) ，but also can N lower the number of vehicle routing problems to be solved from 2 - 1 to 2N - 1. As a result，the cost alloca- tion of collaborative delivery can be completed within a reasonable time. Calculation results of numerical ex- amples and actual cases show that the time needed for the B-T Shapley method is almost negligible compared with that for the Shapley value method. Moreover，the cost allocation results using the B-T Shapley method only have minor deviations compared with those using the Shapley value method，with an average accuracy of a- bout 95% .