The calibration of the engine control unit is increased for the development of the whole automotive system. The aim is to calibrate the electronic engine control to match the decreasing emission requirements and increasing fuel economy demands. The reduction of the number of tests on vehicles represents one of the most important requirements for increasing efficiency of the engine calibration process. However, the definition of the design of experiment is not straightforward because the data is not known beforehand, so it is difficult to process and analyse this data to achieve a globally valid model. To reduce time effort and costs the virtual calibration can be a valid solution. This procedure is called software in the loop (SIL) calibration able to develop a process to systematically identify the optimal balance of engine performance, emissions and fuel economy. In this work, a virtual calibration methodology is presented by using a two-stage model to get minimum exhaust emissions of a diesel engine. The data used are from a GT-Power model of a 3L supercharged diesel engine. The model is able to calculate the engine emissions for different engine parameters (such as the start of injection, EGR fraction and rail pressure) and from optimisation process, new injection start maps that reduce pollutant emissions are created.