Soil response to contamination with 2,4,5-triclorophenol was studied to test the validity of the concept of Generic Reference Levels (GRL), the main criterion used to define soil contamination. Soil samples were artificially contaminated with doses between 0 and 5000 mg kg(-1) of 2,4,5-triclorophenol, and analysed by various tests. Where possible, the response of soils to the contaminant was modelled by a sigmoidal dose-response curve in order to estimate the ED(50) values. The tests provided different responses, but only microbial biomass-C and dehydrogenase and urease activities demonstrated soil deterioration in response to contamination. The results suggest that the diagnosis of soil contamination has been greatly simplified in the legislation by the provision of a single figure for each compound, and that the GRL concept could perhaps be substituted by measurement of ED(50) values, which better reflect the alteration of a soil due to the presence of a xenobiotic substance.