The purpose of this paper is to present a global description of an efficient, robust and clearly structured control design, based on different hierarchical predictive algorithms, applied to the automotive engine management. The main advantages of these structures are to be well adapted to engines with multiple combustion modes: homogeneous or stratified modes for gasoline direct injection engines, gasoline or compressed natural gas for hybrid fuels engines, etc…A global hierarchical structure is presented, with different levels of acting, from local regulations to global coordination. A model based predictive controller is recommended, to reduce the cross-influences in dynamics and provide a separation of each parameter adjustments. The global performances are thus not dependent on the numerical precision but on the structure quality of the controllers. The more complex the automotive systems become, the more structured the controllers must be designed.