The claim that models are representationally inadequate, as the title of this special issue tentatively suggests, is provocative. Isn't it the case that, by their very nature, models aim at idealization, approximation, and simplification? These features are often seen as merits rather than defects of models. Pragmatists and instrumentalists have argued extensively that this kind of "inadequacy" does not matter, as long as models serve their descriptive or predictive purposes. However, models also seem to play a vital role in understanding and explaining reality and in giving us descriptions of what there is; prima facie, their function does not reduce to merely enabling us to somehow get along. Given their representational "deficiencies", it is not at all clear to which extent and how models can help us understand the world, or how they can possibly exhibit something like explanatory power.In recent years, various proposals concerning the metaphysics of models as well as their representational nature have been advanced (Frigg 2010;Godfrey-Smith 2006;Weisberg 2013;Giere 2004;Alexandrova 2008;Toon 2012; Bokulich 2009;Strevens 2008;García-Carpintero 2010). However, these accounts have not provided a satisfactory or broadly accepted answer to the epistemic value of representationally inadequate models. The pragmatic value of at least some scientific models is beyond suspicion, but do they also enable us to better understand and explain their target systems? More specifically, it is not clear what are the features in virtue of which some (inadequate) models provide understanding, while others do not.