In times where the attention on alternative energy sources is continuously increasing, the study of biofuels is taking a primary role, as a possible replacement, or integration, of traditional transportation fuels. Among them, biodiesel fuels are typically a complex mixture of large fatty acids, obtained through trans- esterification of soybean and rapeseed oils with methanol. In this background, the pyrolysis and combustion kinetics of methyl esters is essential for a proper understanding of the combustion behavior and pollutant formation from biodiesel fuels. From a modeling point of view, when studying the combustion of large methyl esters a reliable kinetic mechanism is required. Nevertheless, the main problems of these molecules lie in their length and lack of symmetry in their structure, which results in the huge size of the related detailed kinetic mechanisms (thousands of species) with a consequent significant computational load, even when dealing with 0D and 1D models. In order to maintain the applicability of the kinetic mechanism also in multidimensional models, it should be useful someway to reduce its dimensions. For this reason, a reduced kinetic scheme for methyl esters was developed and is presented in this work. It is the result of the coupling of two different techniques: (i) an upstream lumping of species and reactions, through which several species are grouped into a single pseudo-species according to proper rules; (ii) a successive further reduction of the kinetic mechanism through a novel technique, based on the analysis of the reacting system in the desired range of operating conditions. The reduced mechanism was validated through comparison with experimental data in a wide range of conditions using shock tube, laminar flame speeds and ideal reactors. A satisfactory agreement with both experimental data and the original scheme was observed. Additionally, thanks to its limited dimensions, the kinetic model could be applied on more complex, multidimensional models. As an instance, its performances on a multi-zone model of an HCCI (Homogeneous Charge Compression Ignition) engine are assessed: the obtained results show the great advantages of this mechanism, which can then be used in place of the original one without losing in accuracy, but with considerable savings in computational times.