In the last decades there has been a significant increase of traffic demand in urban areas due to the development of the economy and the increase of population. The increase in traffic demand leads to a more congested road network. Congestion causes unwanted delays, resulting in higher travel costs, noise, and pollution. Popular methods to resolve the problem of congestion are, for instance, to improve the use of the existing infrastructure or to extend the current infrastructure. Most of the time the latter is not possible, because there is no space for extension of road network. To improve the use of the existing infrastructure, the available capacity should be utilized as efficiently as possible. In an urban road network traffic travels from intersection to intersection. Hence, the intersections are influenced by each other with some time delay, and therefore it would be useful to predict the evolution of the traffic demands. For this reason an Model Predictive Control (MPC) strategy will be suitable for controlling traffic. However, this is to realize in practice due to the long computation time and the presence of uncertainty in traffic. The goal of this thesis is to develop a predictive model-based urban traffic controller that accounts for uncertainty while not losing performance in every traffic regime and while remaining real-time feasible. The aim of the controller is to improve the throughput of an urban traffic network by aggregating the traffic dynamics to (several) tens of seconds, and this is evaluated by means of simulation.