Abstract: At present, most of the research on green vehicle routing problem focus on the deterministic environment, in which distance and vehicle speed are considered as the influencing factors of carbon emissions, and the objectives generally consist of various logistics costs, including fuel costs and vehicle using costs. In this paper, the green vehicle routing problem with fuzzy demand and soft time window was proposed, in which the vehicle speed and gross weight were considered as the influence factors of the carbon emissions, and a multi-objective programming model was established to not only minimize fuel costs, carbon transaction costs, vehicle using costs, but also maximize customer satisfaction. The model was solved by wolf pack algorithm, in which the walking behavior, calling behavior, siege behavior and wolf swarm update mechanism was defined. During the optimization of the algorithm, the "real value" of each objective was calculated by stochastic simulation algorithm, which was embedded into the wolf pack algorithm as the fitness of artificial wolf. The effectiveness of the algorithm was analyzed and the algorithm was compared with other algorithms by using the simulation experimental data. The results show that the algorithm is feasible and effective. This paper also analyzed the influence of the change of decision-maker’s subjective preference value Cr* on each goal and the interaction between each goal.