A variant of vehicle routing problem in which customers require simultaneous pickup and delivery goods during specific individual time windows (VRPSPDTW) is addressed. A general mixed integer programming model focused on minimizing the number of vehicles is proposed. A simulated annealing (SA) algorithm is developed and applied to solve this NP-hard optimization problem. Computational results are reported for a set of 15 test problems from Wang and Chen’s benchmark and compared with the results from a genetic algorithm (GA). It is shown that SA is an effective metaheuristic to solve the VRPSPDTW.