The use of mathematical models derived from physical principles is gaining more widespread acceptance for automotive control and diagnostic applications. A suitable mathematical model may reduce, though not eliminate, the need for empirical calibrations, and may help in accommodating changes in operating conditions, external disturbances, vehicle to vehicle variability, aging etc. Recent studies have shown that model based approaches for both control and diagnostic design offer a viable alternative to empirical methods for industrial applications. However, until recently, model-based control and diagnostic algorithms have been designed separately, without considering their interactions explicitly. As a consequence, the performance of these algorithms may be limited, and even deteriorated in the presence of modeling uncertainty and disturbance. To overcome this difficulty, a method for the systematic integration of control and diagnostic modules is proposed in this paper, and is applied to the control and diagnosis of air and fuel dynamics in an internal combustion engine. Some experimental and simulation results will be included to illustrate the possible advantages of the proposed method.