Soils from urban and peri-industrial regions are affected by the presence of technogenic susbstrates and the fallout of smoke and dust dispersed by the winds in the vicinity of pollution sources. Pollutant levels in these soils display significantly higher variability than in soils that remained close to natural state. Identification of the factor controlling this variability is the first step toward prediction and mapping of pollutant levels in such soils. This work is aimed at characterizing the distribution of trace metal concentrations in urban and peri-industrial soils from the Walloon region (Belgium) through a descriptive model. We selected 8 areas representating different current and historical industrial activities of the region such as chemical plants, steel industry (blast furnaces, steel), coal extraction (extraction and agglomeration of coal, coke and coal chemical) and non-ferrous metallurgy (Pb, Zn). The sampling points (70 to 200 in each area) were spatially well distributed and were representative of the various land uses in peri-industrial area: gardens, parks, farmland and forest. Soils with suspected local pollution (eg.: dumps or industrial site) were not selected. At each point, we collected a topsoil sample corresponding to the first soil horizon and, when possible, a deeper horizon (between 60 cm and 1 m depth) was taken with a hydraulically powered percussion hammer. On each sample, pseudo-total concentrations (aqua regia) were measured for Al, As, Ba, Be, Ca, Cd, Co, Cr, Cu, Fe, Hg, Mg, Mn, Mo, Ni, Pb, Se, Sn, Sb, Tl and Zn. Our soil typology was build with different statistical and graphical tools (boxplot, scatterplot, PCA). For highlighting soils categories with a clear signal in terms of trace metal content, we use robust statistics including robust methods for outlier identification. The typology is based on four main factors: (1) the geopedological substrate, (2) the industrial context, (3) the amount and nature of technogenic materials and (4) the land use. This typology ...