Abstract: In order to promote distribution service and reduce delivery cost of tobacco logistics, a carbon emission mathematical model on a background of low carbon was established by introducing workload equilibrium index, and an improved double-layer genetic algorithm was proposed to optimize delivery route. Taking workload balancing as a target and by way of clustering, the first layer converted an issue of multi-vehicle to multi-service point into an issue of single-vehicle to multi-service point. In the second layer, a tabu genetic algorithm was designed to promote the precision of model through adding a memory function to the tabu list. Taking the distribution center of Zhejiang Provincial Tobacco Corporation as an object, the said algorithm and the model were validated. The results showed that the algorithm was superior to single-layer genetic algorithm in terms of calculation time and convergence effect. The optimized delivery route decreased the delivery cost by 25.4%, increased average wholesale efficiency by 36.8%, and raised average transfer efficiency by 33.9%. The algorithm and the model provide technical supports for the optimization of delivery route in tobacco logistics.