Liquid injection optimization is key to enhance the efficiency of many processes in a wide variety of fields. Combustion processes are one of the most challenging ones, due to the direct emissions and greenhouse gases implications. This work aims at studying the external two-phase flow produced when injecting fuel with a pressure-swirl atomizer in an operating condition typical of an academic burner. The main objective is to have a better comprehension of the atomization process in this type of nozzles, getting information on the droplets’ characteristic size and how they are produced and arranged. For that regard, a high-fidelity Direct Numerical Simulation (DNS) is used to analyse the very near field of the spray produced, where primary atomization takes place. Pre-processing tasks are explained in terms of geometry comprehension, computational domain and mesh selection, boundary conditions and main simulation setup parameters. Then, through post-processing tasks, both quantitative and qualitative results are extracted, which will serve to validate the modelling against previous works and to provide novel data about the atomization process in pressure-swirl atomizers. Results show that smaller droplets predominate over bigger ones as expected, since just the first millimetres of the spray are modelled. However, there is a clear trend of droplet’s size growing when increasing both axial and radial distances, indicating coalescence in regions relatively far from the nozzle. The achieved results, together with results from simulating the injection event with other fuels or at other operating conditions, can be used to develop a phenomenological model able to predict how atomization is going to be as a function of the non-dimensional Reynolds and Weber numbers that could be implemented in lower-resolution RANS and LES codes for modelling atomization. This investigation proves that it is possible to faithfully predict the near field of these sprays through DNS simulation, getting similar trends to those of the experimental data, and that the study through numerical models is necessary in the investigation process. The information they can bring, together with the experimental knowledge, can make a good synergy that will eventually lead to a better understanding of this type of atomizers.