The Eurostat `nomenclature of territorial units for statistics' (or NUTS), the three-level hierarchy designed for the regional statistics of EU-27, EFTA, and candidate countries, has both analytical and political purposes: its implementation, favouring existing institutional divisions, resulted in important differences between countries, owing as much to national status and history as to active negotiations/reinterpretations (Eurostat, 2007). The visualization offered here helps to get some insight on this process, with the adaptation of a `circle packing' layout, first proposed by Wang et al (2006) and implemented by Bostock and Heer (2009). Built as an iterative process starting from the upper level of the hierarchy, this layout aims at compactness (sorting circles according to their size, then placing each tangentially to the circle that is nearest to the centre of the layout) and readability (in case of overlaps, circles are deleted, swapped, and redrawn). As in the present graphic, the final sorting of entities results partly from trial and error, and is close to, but not fully congruent with the ascending order. We emphasize Eurostat's lower-level regions (NUTS3), depicted with solid discs sized proportionally to their actual geographical area. The layout includes the overarching hierarchy, with higher NUTS levels (2, 1, and 0, that is, countries) denoted with nested circles (dashed lines and lightened tones). Overestimating the area of upper levels, the display assumes a trade-off between structure legibility and areal accuracy and allows us to convey additional information, which is illustrated here with a population density threshold (150 inhabitants/km2, a rough proxy of the rural/urban nature of the region). Similarities in national patterns -for example, between Scandinavian countries (SE, NO, and FI), are revealed by visual inspection, as is the skewness of the distribution of NUTS3 regions in Germany (DE) and the United Kingdom (UK), mixing large rural regions with numerous smaller urban entities. Other countries (FR, RO, BG), appear to have favoured spatial evenness and intralevel consistency over functional and population-centered logics. The layout also emphasizes how a density-dependent category can change along the levels of the hierarchy.