The constant increase in pollutant emissions constraints has obliged automotive manufacturers to adopt a global optimization approach of engine and exhaust after-treatment technology. Engine control strategies appear to be a powerful solution to address this issue. The problem is particularly complex since acceptable drivability must be maintained whilst at the same time reducing in-cylinder pollutant emissions and ensuring optimum conditions to attain highconversion efficiencies via exhaust gas after-treatment systems. The development of appropriate control strategies can only be achieved with an in-depth understanding of the engine behaviour, using experimental results and system numerical simulations. In this context, predictive combustion and pollutant emissions models, which are calibrated with experimental data, are particularly useful as they allow a wide range of parametric variations to be studied. This paper presents an advanced Diesel combustion model based on a Barba's approach [Barba C. et al. (2000)- A Phenomenological Combustion Model for Heat Release Rate Prediction in High Speed DI Diesel Engines with Common Rail Injection , SAE Technical Paper 2000-01-2933]. This model can be applied to multi-injection, defining a pre-mixed combustion zone for the pilot injection and a diffusion combustion mode for the main injection. To assess the in-cylinder pollutant emissions, a mixing model based on the turbulent kinetic energy generated by the spray, is added to define a burnt gas zone in which post-flame chemistry including CO, NOx and soot formation can be computed. This model is first validated using CHEMKIN and 3D CFD results. Then, using experimental results, a 4 cylinders D.I. Diesel engine is calibrated on steady state engine operating conditions and coupled to an engine control to predict the evolution of pollutant emissions under transient conditions.