In recent years there has been growing interest in algorithms inspired by the observation of natural phenomena to define computational procedures which can solve complex problems. In this paper, through an analysis of the constructive procedure of the solution in the ant colony system (ACS), a vehicle routing problem (VRP) is examined and a hybrid ant colony system coupled with a stochastic local search algorithm(SLSACS), is proposed. In SLSACS, only partial customers are randomly chosen to compute the transition probability. Experiments on various aspects of the algorithm and computational results for fourteen benchmark problems are reported. We compare our approach with ACS, some other classic, powerful meta-heuristics and show that our results are competitive.