The fuel injection system of diesel engines is of great importance since it controls the combustion mechanism. The rate of injection and the speed of injected fuel are important parameters for engine operation, controlling the combustion and pollutants formation mechanisms. A fuel injection system simulation capable of predicting the performance of the injection system to a good degree of accuracy has been developed. The simulation is based on a detailed geometrical description of the injection system and in modeling each subsystem as a separate control volume. The simulation starts at the driving mechanism of the fuel pump and describes all parts of the system pump chamber, delivery valve, delivery chamber, connecting pipe and injector. The components of the system are put together and interact as they do in reality. From the cam geometry an analytical expression is derived that gives the pump piston lift as a function of the engine crank angle. The equations of continuity and momentum are solved using the method of characteristics inside the pump chamber using a constantly moving mesh with boundary conditions derived from the motion of the plunger, while up to now most researchers considered the pressure inside the pump chamber uniform. The delivery valve and injector needle movement is obtained using Newton's second law by taking into consideration the various forces acting upon them. A second set of characteristics is used to solve the unsteady flow inside the delivery valve chamber and the connecting pipe. The fuel is considered to be compressible at every point of the injection system except from the injector needle seat and the resulting system of the two equations (continuity and momentum) is solved for speed and pressure. The prediction of the pressure-time diagram at any point of the injection system is possible. For the validation of the model, experimental data have been taken from a diesel engine operating at various speeds and loads. The fuel pressure in the line between the pump and injector has been recorded using a pressure transducer and a digital computer at a rate of 5 points per degree of crank angle and compared to the one predicted by the model. A good match between predicted and experimental pressure traces has been obtained revealing the predictive ability of the simulation model.