Feed-forward low-throughput models have been developed to predict MFB50 and to control SOI in order to achieve a specific MFB50 target for diesel engines. The models have been assessed on a GMPT-E Euro 5 diesel engine, installed at the dynamic test bench at ICEAL-PT (Internal Combustion Engine Advanced Laboratory at the Politecnico di Torino) and applied to both steady state and transient engine operating conditions.MFB50 indicates the crank angle at which 50% of the fuel mass fraction has burned, and is currently used extensively in control algorithms to optimize combustion phasing in diesel engines in real-time. MFB50 is generally used in closed-loop combustion control applications, where it is calculated by the engine control unit, cycle-by-cycle and cylinder by-cylinder, on the basis of the measured in-cylinder pressure trace, and is adjusted in order to reduce the fuel consumption, combustion noise and engine-out emissions.A feed-forward approach has been developed in this paper. This approach is capable of predicting MFB50 on the basis of several parameters, such as, the in-chamber thermodynamic conditions, the injected fuel quantities and timings, the injection pressure, the oxygen concentration and the engine speed and load. The approach has also been inverted in order to predict the start of injection required to achieve a specific MFB50 target in real time. This method can be used in model-based real-time control algorithms to adjust the engine parameters in order to prevent the occurrence of non-optimal combustion cycles.