Iron-catalysts, based on ZSM-5 (FeZSM5) and Cuban natural Mordenite (FeMORD) zeolites have been prepared by a conventional ion-exchange method and their catalytic activity in the selective catalytic reduction (SCR) of NO with ammonia was studied in the presence of H2O and SO2. A commercial SCR catalyst (CATCO) based on V2O5–WO3–TiO2, was also studied as a reference. This paper presents the experimental results of using these catalysts without toxic vanadium and also exploits a neural network-based approach to predict NOx conversion efficiency of three SCR catalysts. The mathematical functions derived have been integrated into a numerical model to simulate diesel road vehicles equipped with SCR catalysts such as those studied here. The main results indicate that despite toxic vanadium and N2O formation, CATCO shows better NOx conversion efficiencies. However, FeMORD does not produce N2O and performs better than the FeZSM5. The simulation results on real cycles show lower level of NOx for heavy-duty and light-duty diesel vehicles compared with homologation load cycles.