In order to improve the forecasting accuracy of logistics demand, this paper puts forward a logistics demand forecasting model based on least support vector machines optimized by ant colony optimization algorithm (ACO-LSSVM). Firstly, the data of logistics demand are reconstructed, and then the complex nonlinear change rule of logistics demand is explained through LSSVM, and the parameters of LSSVM model are optimized by ACO, and lastly, the performance of mode are tested by logistics demand data. The simulation results show that ACO-LSSVM has improved the forecasting accuracy of logistics demand, and which is an effective method for logistics demand forecasting.