New French legislation for the management of potentially polluted sites requires that “the state of the environment” (i.e. of the soils and waters of the industrial site under study) be compared to the state of neighbouring natural environments or to regulatory management values established by public authorities. Within the framework of urban planning projects or conversion of former industrial sites, many studies are daily carried out in order to assess the risk to the population. Departments responsible for public health are recipients of these and must put forward an opinion. In the absence of statutory values, everyone tries to find local or regional data relating to trace metal concentrations in the soil (often called ‘background values’), to use as points of comparison and thereby estimate the contamination level of the soil under study. Unfortunately, in such an industrial or urban context, it is very difficult to find “natural soils” located in the vicinity, which are similar enough from a pedological standpoint. More often than not, these assessments can at best be carried out with reference to soils considered to be relatively little-contaminated, namely the closest agricultural soils. In France, a large data collection was carried out on a national scale (known as BDETM 2009). These data were provided by analyses, undertaken mainly within the framework of sewage sludge spreading plans and of some research programmes. The samples were taken from surface horizons of ploughed soils. These analyses only involved Cd, Cr, Cu, Hg, Ni, Pb and Zn and were taken from more than 73,000 sites throughout France. These are the only data available in large numbers. Due to their origins, (topsoil analyses from agricultural soils of various types), these analyses very largely correspond to “usual agricultural contents” i.e. natural pedogeochemical concentration + local additional contamination by farming practices and diffuse atmospheric deposition. These “usual agricultural contents” correspond to the lowest concentrations that it is possible to measure in an urban or peri-urban soil, close to an industrial site. Beforehand different questions arise : on which territories statistics must be established, where and how thresholds must be fixed ? The answers are to be given necessarily at the local level because the same high concentration observed in a soil could be natural in one particular case and anthropogenic in another. The data collected on the area of the region Centre for 5 913 sites have been processed and stratified according to “agricultural districts”, territories much less heterogeneous from geological and pedological standpoints than administrative subdivisions. The usual measures of position (percentiles) and of dispersion (upper whiskers) have been determined for each agricultural district. This collection provides objective points of comparison for any institution or consultancy office wishing to confront the results of an analysis with various statistical indicators to decide whether the measure is anomalous or not. If the available data are sufficiently numerous, these statistics generated stratifying by agricultural districts are the most pertinent. Finally, the diagnosis as to whether a value is truly natural or anthropogenic (hence different risk levels) will have to be based on a pedologically and geologically thorough point study of the site, without forgetting a historical inventory of possible sources of pollution.