This paper describes recent progress towards Direct Injection Diesel engine combustion simulation, involving both engine data measurements and 3D computational fluid dynamics. Experimental data were obtained in two different engines and it includes LDV measurements as well as an extended engine combustion data base. CFD simulations were performed with a modified version of the Kiva2 code and experimental data were used to get initial and boundary conditions and to validate the numerical results.The initial version of the Kiva2 code was enhanced to allow computation of quite complex geometries with different physical sub-models for turbulence, injection and combustion. The present version of the code can compute the engine cycle from start of intake stroke to the end of combustion with any inflow and wall boundary conditions. Turbulence may be simulated using both a k-ε model and a second order uiuj-ε model which was recently implemented in the code. The spray model was also modified in order to limitate the deficencies of current spray models when computing high injection pressure Diesel sprays. At last, two different combustion models are available and were tested against experimental data: the Magnussen combustion model and a 3 equation flamelet combustion model. In all the calculations, initial and boundary conditions were carefully specified using experimental data. In particular the influence of numerical parameters and intake conditions in the calculations were assessed. Numerical results were compared to experimental data from two different DI Diesel engines. First of all LDV measurements were performed in the bowl of an optically accessible engine. It was found that the predicted mean and turbulent velocities are in good agreement with experimental data. Calculated velocities are slightly underpredicted but the computations show the same trends as the experiments with engine RPM. The combustion engine data base includes pressure profiles and combustion analysis in a second engine for a wide range of engine parameters such as RPM, load, injection timing. Provided that the injection conditions are precisely specified, a fair agreement between experimental and numerical results was obtained.As a conclusion, the modified version of the Kiva2 code was assessed in this paper for DI Diesel engine combustion calculations. The results are quite promising and some modeling aspects which need further insvestigation were clearly identified.