This paper addresses the modeling and control of the air path system of diesel engines. The underlying issues are critical for the control of the transient exhaust gas fraction pumped into the cylinders, which is known to be a dominant factor to reduce the nitrogen oxides (NOx) emissions. In this paper, we propose a new approach, based on a data-based grey-box linear parameter varying (LPV) model as well as on the gain scheduled H∞ technique for the controller design. The modeling step is shown to lead naturally to a so-called quasi-LPV structure, which also delivers the scheduling variables to be accounted for. Using this information, gain scheduled H∞ techniques allow to design a controller which enforces a much better tracking performance than the standard production electronic control unit, while not requiring any calibration work. The performance of the proposed approach is demonstrated by experimental results. © 2007 IEEE.