Persistent environmental transformation products are increasingly being detected in surface waters and previous parts of this volume have discussed methods for prediction and quantification. However, there is not sufficient experimental data on their ecotoxicological potential to assess the risk associated with transformation products, even if their occurrence and abundance is known. Herein, we review computational methods for the identification and prioritization of transformation products according to their ecotoxicological potential and specifically focus on the assessment of mixtures of organic environmental pollutants and their transformation products. These transformation products can be produced through abiotic or microbial degradation or from metabolism in higher organisms. The proposed model assumes concentration addition between the components of the mixture and uses Quantitative Structure Activity Relationships (QSARs) to fill data gaps. The model is illustrated for five pesticides and their environmental transformation products. Their overall toxic potential is derived by scaling predicted relative aquatic concentrations (RAC, see Fenner et al., 2008, in this volume) with the relative potencies of each transformation product followed by summing up the toxic potentials of all mixture components. The model is versatile and can also be used to assess the cocktail of metabolites that is excreted by humans and animals after consumption/ingestion of pharmaceuticals. The metabolites of pharmaceuticals and hormones that are excreted are often more hydrophilic and consequently presumably less toxic than the ingested parent compound. However, they may be more abundant and therefore may be relevant for overall risk assessment. The weak point of our method, as of any QSAR application, is the correct assignment of the mode of toxic action (moa) of transformation products because they do not necessarily exhibit the same moa as the parent compound. In the future, more emphasis must therefore be placed on this issue, e.g., by identifying toxicophores or other structural alerts that are indicative of a certain mode of toxic action. An improved mode of action assignment would make the model more robust. Nevertheless, the prediction method is valuable for screening purposes and for setting priorities for further experimental testing. [ABSTRACT FROM AUTHOR]