Though very important for the system performance, the dynamic behavior of the catalytic converter has mainly been neglected in the design of exhaust emission control systems. Since the major dynamic effects stem from the oxygen storage capabilities of the catalytic converter, a novel model-based control scheme, with the explicit control of the converter's oxygen storage level is proposed. The controlled variable cannot be measured, so it has to be predicted by an on-line running model (inferential sensor). The model accuracy and adaptability are therefore crucial. A simple algorithm for the model parameter identification is developed. All tests are performed on a previously developed first principle model of the catalytic converter so that the controller effectiveness and performance can clearly be observed.