Nowadays it is commonly accepted that natural organic matter (NOM) plays a key role as a complexing agent for trace metals in natural systems (surface and groundwaters, soils). However, significant limitations are imposed on the measurement and interpretation of the complexation equilibria where NOM participates because of the fact that it is ill-defined in nature. As a consequence, a real mix of NOM binding models can be found in the literature. The purpose of this article is to discuss how one of the main features of NOM, the heterogeneity of their binding sites, is taken into account in existing discrete models. A better understanding of this aspect should help the potential user to decide which strategy could be used to solve a particular problem.