Abstract. The aim of this paper is to understand resonance production (and more generally particle production) for different collision systems, namely proton-proton (pp), proton-nucleus (pA), and nucleus-nucleus (AA) scattering at the LHC. We will investigate in particular particle yields and ratios versus multiplicity, using the same multiplicity definition for the three different systems, in order to analyse in a compact way the evolution of particle production with the system size and the origin of a very different system size dependence of the different particles.The main goal of heavy ion physics at very high energies is the proof of existence of the creation of a quark-gluon plasma, and the study of the properties of this exotic state, by analysing the final state of many thousands of produced hadrons. Unfortunately, many of these final state particles are not directly coming from the decay of the plasma, but they are produced or at least affected by the hadronic cascade, the last stage of the collision before particles freeze out and freely move to the detectors. Here resonance studies come into play : There is a large number of resonances at our disposal, having very different lifetimes, from 1 fm/c to several tens of fm/c, which means that these particles decay with varying probabilities in the hadronic stage, and provide therefore valuable information about the latter one.We extended this analysis, to also consider small collision systems (pp and pA), in addition to AA. We also consider not only resonances, but also stable particles, providing useful additional information (the lifetime is not the only relevant parameter). Our tool to analyse particle production is the EPOS model. EPOS3 [1] is a universal model in the sense that for pp, pA, and AA collisions, the same procedure applies, based on several elements: