This paper presents a new approach of Air Fuel Control (AFR) control strategy based on non-linear dynamics models of the engine air and fuel processes. Higher demands for lower engine exhaust emissions require a more accurate control of the air-fuel ratio during transient operations. A model-based air-fuel ratio control was applied to a production multipoint port injection engine with the initial constraint to keep the original sensors and actuators. The control strategy contains real-time engine models which describe air, fuel and sensors dynamics. In the final step, the control structure was implemented on a production car and evaluated on a chassis dynamometer. Several experiments with different transient conditions have shown reduced air-fuel ratio excursions in magnitude and duration compared to the standard strategy. Moreover, the calibration process was considerably easier than usual and required less testing on the car.