A modified version of KIVA II code was used to perform three-dimensional calculations of combustion in a DI diesel engine. Both an ignition delay submodel and a different formulation of the fuel reaction rate were implemented and tested.The experiments were carried out on a single cylinder D.I. diesel of 0.75 I displacement equipped with sensors to detect injection characteristics and indicated pressure. A fast acting sampling valve was also installed in the combustion chamber to allow the measurement of main pollutants during the combustion cycle, by an ensemble average technique. Computational and experimental results are compared and the discrepancies are discussed.Today the demand for light duty engines that produce less emission and consume less fuel is increasing. Thus, if limits on CO2 emissions are established, the direct injection diesel engine for light duty applications will become an attractive option. Also, manufacturers of heavy duty diesel engines are facing increasingly stringent emission standards. New research efforts aim at improving DI diesels emission behaviour.Given the complexity of the interaction between the different processes in diesel combustion, comprehensive models of in-cylinder phenomena prove to be useful in providing guidelines for prototype development. Multidimensional analytical models of in cylinder phenomena have been greatly improved over the last decade, given the progress of the efficiency of computer hardware and diagnostic techniques. In particular many authors [1, 2, 3, 4, 5, 6 and 7] have numerically simulated the interaction between the airflow field and the spray and have compared it with experimental data. However, both three dimensional calculations of combustion in the presence of fuel sprays and the corresponding comparisons with experimental data are hardly found in literature. In most cases the comparison between numerical and experimental results is limited to the pressure history and to the fitting of some pollutant emissions late in the expansion stroke.In particular, limiting the review to diesel combustion, Zellat and others  present calculations of a swirl chamber engine at different loads, injection timings, and speeds, obtained with a modified version of KIVA code. In their results, the computed in-cylinder pressure agrees well with the experimental one. Takeneka and others  used KIVA code to perform calculations of DI Diesel combustion for both swirl chamber and quiescent chamber systems. No details about the modifications introduced in the original code are provided.Computational comparisons with in cylinder measurements, performed in a quiescent chamber system by Kamimoto , show discrepancies between computations and measurements. Gonzales and others  also used the KIVA II code with the Reitz atomization model and a single step Arrhenius mechanism to describe combustion of tetradecane in a D.I. quiescent chamber engine. The paper focuses on the spray submodel effect on the accuracy of pressure history prediction. It was demonstrated that, if adequate grid resolution is used, prediction of combustion is sensitive to the details of combustion and fuel injection models.In the present paper a detailed comparison between experiments with a fast acting sampling valve and 3D calculations is presented for a medium duty swirl supported DI engine. Calculations are performed by KI-VA II code with the adoption of the TAB jet break up sub-model and wall impingement modelling. In addition, a different formulation of combustion model is discussed. The current status of the “numerical engine program” in progress at Istituto Motori of CNR, is thus defined.