Abstract: To optimize tobacco logistics delivery process, by taking the minimization of the distance of delivery amount and the variance of delivery time, and the maximization of the difference between service areas as indexes, an integrated optimization model for logistics system was established via weighting and mixing. Using the solution generated by clustering algorithm as the initial elitist-population in genetic algorithm, and using the self-adaptive mutational operator based on the variance of delivery time to modify the traditional genetic algorithm, a self-adaptive elitist-based genetic clustering algorithm was designed to accelerate the convergence of cluster and even the distribution of cluster areas. The designed algorithm was verified in a city level tobacco logistics delivery center in Zhejiang Province, the results showed that:1) After the optimization of three transit stations and a delivery center, the direct delivery rate to retailers increased by 7.4 percentage points, the direct delivery rate by volume increased by 2.9 percentage points, and total delivery mileage reduced by 1 352 km. 2) The logistics, sorting and storage costs per 10 000 cigarettes decreased significantly and were lower than their respective averages in Zhejiang Province. 3) The major cigarette markets in the region agreed with the delivery areas planned by the designed algorithm. This method is practicable and provides technology support for promoting the efficiency of logistics delivery.