The progress of soil clean up in the Netherlands is severely hindered by the lack of common agreement on how to describe the grey areas of increased background levels of pollutants. In this study practical methods are proposed in which background levels are described as distribution functions within map units of comparable homogeneity. To stimulate large scale implementation the statistical methods are simple and robust. In most applications percentiles are preferred above averages. Applications include the trigger value for further investigation and the target values for soil clean up. Special attention has been given in developing methods which avoid high sampling costs. Aspecially the described method for evaluation of soils can be re-used can lead to considerable cost reduction, while maintaining a soil protection policy of 'stand still'.