Gasoline direct injection (GDI) allows flexible operation of spark ignition engines for reduced fuel consumption and low pollutants emissions. The choice of the best combination of the different parameters that affect the energy conversion process and the environmental impact of a given engine may either resort to experimental characterizations or to computational fluid dynamics (CFD). Under this perspective, present work is aimed at discussing the assessment of a CFD-optimization (CFD-O) procedure for the highest performance of a GDI engine operated lean under both single and double injection strategies realized during compression.An experimental characterization of a 4-stroke 4-cylinder optically accessible engine, working stratified lean under single injection, is first carried out to collect a set of data necessary for the validation of a properly developed 3D engine model. Homogeneous lean operation is not considered due to the consequent high instability of the engine under study as injection is too much advanced. The 3D engine model is used to explore the main advantages deriving from modulating injection on the engine power output and the main pollutants formation. The model is then exploited to perform an optimization analysis searching for the injection and spark advance synchronization within the working cycle being optimal for the combustion development.Numerical simulations allow clarifying the effects of multiple injections on the engine power output and the NO and CO emissions, besides offering the way to study the combustion process and the occurrence of the adverse phasing deriving from variations of the spark timing. The obtained results are finally used to drive a further experimental campaign that substantially confirms the main conclusions of the CFD-O, at least qualitatively.