The problems associated with vehicular traffic are becoming more noticeable in large cities and highways, mainly because it is becoming less feasible to build new roads or expand the existing ones. This issue is beginning to be treated as a control systems optimization problem, where the objective is to maximize the use of the existing infrastructure. However, the size of the traffic networks and the large number of variables make the implementation of optimization-based centralized control schemes not feasible. This article presents the development of distributed and decentralized predictive controllers for traffic control purposes. The proposed schemes are tested in simulation environment using a highway with two access ramps and an urban traffic network composed by three intersections.