The combination of more strict regulation for pollutant and CO2 emissions and the new testing cycles, covering a wider range of transient conditions, makes very interesting the development of predictive tools for engine design and pre-calibration. This paper describes a new integrated Virtual Engine Model (VEMOD) that has been developed as a standalone tool to simulate new standard testing cycles. The VEMOD is based on a wave-action model that carries out the thermo-and fluid dynamics calculation of the gas in each part of the engine. In the model, the engine is represented by means of 1D ducts, while the volumes, such as cylinders and reservoirs, are considered as 0D elements. Different sub-models are included in the VEMOD to take into account all the relevant phenomena. Thus, the combustion process is calculated by the Apparent Combustion Time (ACT) 1D model, responsible for the prediction of the rate of heat release and NOx formation. Experimental correlations are used to determine the rest of pollutants. In order to predict tailpipe pollutant emissions to the ambient, different sub-models have been developed to reproduce the behavior of the aftertreatment devices (DOC and DPF) placed in the exhaust system. Dedicated friction and auxiliaries sub-models allow obtaining the brake power. The turbocharger consists of 0D compressor and turbine sub-models capable of extrapolating the available maps of both devices. The VEMOD includes coolant and lubricant circuits linked, on the one hand, with the engine block and the turbocharger through heat transfer lumped models; and on the other hand with the engine heat exchangers. A control system emulating the ECU along with vehicle and driver sub-models allow completing the engine simulation. The Virtual Engine Model has been validated with experimental tests in a 1.6 L Diesel engine using steady and transient tests in both hot and cold conditions. Engine torque was predicted with a mean error of 3 Nm and an error below 14 Nm for 90 % of the cycle duration. CO2 presented a mean error of 0.04 g/s, while during 80 % of the cycle, error was below 0.44 g/s.