Road transportation sector faces nowadays the challenge of satisfying the growing demands for mobility and at the same time to reduce its negative impact on environment. The panorama for the future is to reduce to zero the tank-to-wheel emissions levels with electric vehicles. Nevertheless, the extended use and massive production of this kind of vehicles is compromised due to different factors. On the one hand, the technology has to keep evolving in order to improve its reliability and to reduce acquisition costs. And on the other hand, cities, legislations and users have to get ready in order to make an appropriate use of these vehicles. What will happen in short and medium term is the transition from current vehicles, powered by internal combustion engines, to 100% electric cars. For this transition, diesel engines will continue having an extended use, or even more, due to their capacity of accomplishing with emissions legislations with low fuel consumption. To be competitive, diesel engine technology has to continue improving in order to satisfying markets requirements being respectful with environment. Cold start is one of the most problematic combustion phases for diesel engines. During this phase, a large proportion of pollutants are produced within the cylinder due to misfiring and incomplete combustion due to the low engine temperatures. Furthermore, exhaust after-treatment devices work inefficiently since their minimum operation temperatures can not be reached. In addition to this, at temperatures below 0ºC, the possibility of continuous misfiring could actually impede the engine start. This is the combustion phase in which this study has been focus on. Different authors have made efforts to give directions in pro of cold start combustion optimization. But these studies are scarce and it is missed a fundamental understanding of how diesel combustion is under this conditions. Knowing this, the planned objective for this thesis is to contribute to the understanding of the cold start combustion process in high speed direct injection diesel engines. This understanding is directed toward the comprehension of the physical and chemical processes that control autoignition and the influence of the different engine parameters. With the purpose of achieving this objective, the first step has been to set-up an experimental facility which allows to reproduce repeatably and systematically cold start conditions. This facility is equipped with an optical engine which operating at room temperature has been adapted to reproduce in-cylinder thermodynamic conditions as those that can be reached during start in a real engine at -20ºC. During the tests, in-cylinder pressure and the main intake and exhaust variables are carefully measured and are later used as input of a thermodynamic model used to calculate the heat release law. Engine measurements are synchronized with a high speed visualization system with high temporal and spatial resolution which allows to "see" how combustion develops within the combustion chamber. All this experimental information has been complemented with the experimental characterization of the injection system and theoretical modeling tools. The methodological thesis approach, the novel experimental facility and the different parametric studies performed have allowed to gain more insight about the cold start combustion process. In first place, a complete phenomenological description of how diesel combustion develops under this conditions have been pieced together. In addition, it has been evidenced, with an important experimental sample, the influence that the different engine parameters have on ignition and combustion progress. And, as one of the major contributions of this work, it has been developed a detailed explanation on how interact the physical and chemical processes that lead to autoignition.