Abstract: This Paper proposed a Multiple Neighborhood Guided Local Search Algorithm (MN_GLS) to solve vehicle routing problem with simultaneous delivery and pickup. Firstly, it used the nearest neighbor method to build the initial solution. Secondly, it did the local search in multi-operator from the initial solution, and found the bow which had the biggest utility of punishment value when the solution fail into the local optimal solution, then changed the punishment of features value and objective function value. Thirdly, it selected the current optimal solution from the local optimal solutions, and then did the local optimization in multi-operator again from the current optimal solution which has the new objective function value. By means of 54 examples, the simulation results illustrate that MN_GLS is an effective and stabilize method for Vehicle Routing Problem with Simultaneous Delivery and Pickup.