Little or nothing is known about the toxicity of most of the >100,000 chemicals released into the environment. The cost of obtaining such information experimentally would be enormous in terms of money, time and animals. Companies and regulatory agencies are therefore turning to the prediction of environmental toxicity and fate through the use of quantitative structure-activity relationships (QSARs). This paper examines the use of QSARs in this context. Generally, QSARs are mechanism-specific, so that the first step is to classify a toxicant into one of four broad classes of toxicity: non-polar narcosis, polar narcosis, unselective reactivity and specific action (e.g. anticholinesterase activity). An appropriate QSAR can then be selected in order to predict the toxicity of a given chemical. There are also some expert systems available for toxicity prediction. QSAR predictions of bioconcentration, soil sorption and biodegradability can also be made; again, expert systems are available for such prediction. QSAR and expert system prediction of physico-chemical properties such as partition coefficient, aqueous solubility, melting and boiling point, vapour pressure and Henry's law constant can readily be made.
QSAR; environmental toxicity; environmental fate; physicochemical properties
