Heavy-duty vehicles will continue to rely on diesel engines due to their many advantages compared to other technologies. However the pollutant emission of these engines are critical even with new advanced fuels. The exhaust gas recirculation (EGR) can improve the emissions, but conventional physical simulation models can be too slow to control efficiently the different EGR modes. Therefore, this work aims to investigate a new approach by using artificial neural networks (ANN) to predict the effect of the control parameters. Measurements with different EGR modes were carried out on a medium-duty commercial diesel engine. The results were used to train ANNs to predict the fuel consumption, exhaust NOx concentration and smoke opacity. The networks were implemented in a simulation and the performance of the models were compared with measured signals. The results will be shown for the NOx emission levels, in simple, transient cycles. At this stage of the research, relative tolerances can be kept below 20%. Therefore the application of artificial intelligence is a promising approach to further improve the emission of diesel engines by advanced control algorithms.