Soil pollution constitutes one of the major threats to public health, where spreading to groundwater is one of several critical aspects. In most internationally adopted frameworks for routine risk assessments of contaminated land, generic models and soil guideline values are cornerstones. In order to protect the groundwater at contaminated sites, a common practice worldwide today is to depart from health risk-based limit concentrations for groundwater, and use generic soil-to-groundwater spreading models to back-calculate corresponding equilibrium levels (concentration limits) in soil, which must not be exceeded at the site. This study presents an extensive survey of how actual soil and groundwater concentrations, compiled for all high-priority contaminated sites in Sweden, relate to the national model for risk management of contaminated sites, with focus on As, Cu, Pb and Zn. Results show that soil metal concentrations, as well as total amounts, constitute a poor basis for assessing groundwater contamination status. The evaluated model was essentially incapable of predicting groundwater contamination (i.e. concentrations above limit values) based on soil data, and erred on the unsafe side in a significant number of cases, with modelled correlations not being conservative enough. Further, the risk of groundwater contamination was almost entirely independent of industry type. In essence, since neither soil contaminant loads nor industry type is conclusive, there is a need for a supportive framework for assessing metal spreading to groundwater accounting for site-specific, geochemical conditions.