With the advent of e-commerce, the market has started to act more and more intensively across its geographic borders and, as a consequence, companies constantly need innovations and improvements in the management of their operations in order to remain competitive. Thus, finding freight solutions that can serve long distances in the short term can be as decisive as the cost factor, for this reason, academic studies in intermodal routing optimization are continually evolving to approach real models. In this context, this thesis seeks to solve a problem of air-land cargo routing, with predetermined airlines and their own heterogeneous fleets. We elaborated an extension of the vehicle routing problem by including arcs that represent overhead lines. The model is applied to a Mixed Integer Linear Programming solver and, firstly, a validation test is performed with fictitious demands in all locations. It is then applied to the actual planning of a government agency in three different periods. We performed analyses on the solution speed; the decision to use the air, land or intermodal modal; and about the earnings of the model. In comparison with the routes actually carried out, the model reduces transport costs by 7 percent to 55 percent. With this results, it is concluded that it is essential that owners of their aircraft and trucks fleet use the air modal only as secondary activities, in other words, that they are fulfilling more missions together (transportation of passengers, for example), or to deliver to remote locations.