The increasing, anthropogenic demand for chemicals has created large environmental problems with repercussions for the health of the environment, especially aquatic ecosystems. As a result, the awareness of the public and decision makers on the risks from chemical pollution has increased over the past half-century, prompting a large number of studies in the field of ecological toxicology (ecotoxicology). However, the majority of ecotoxicological studies are laboratory based, and the few studies extrapolating toxicological effects in the field are limited to local and regional levels. Chemical risk assessment on large spatial scales remains largely unexplored, and therefore, the potential large-scale effects of chemicals may be overlooked. To answer ecotoxicological questions, multidisciplinary approaches that transcend classical chemical and toxicological concepts are required. For instance, the current models for toxicity predictions - which are mainly based on the prediction of toxicity for a single compound and species - can be expanded to simultaneously predict the toxicity for different species and compounds. This can be done by integrating chemical concepts such as the physicochemical properties of the compounds with evolutionary concepts such as the similarity of species. This thesis introduces new, multidisciplinary tools for chemical risk assessments, and presents for the first time a chemical risk assessment on the continental scale. After a brief introduction of the main concepts and objectives of the studies, this thesis starts by presenting a new method for assessing the physiological sensitivity of macroinvertebrate species to heavy metals (Chapter 2). To compare the sensitivity of species to different heavy metals, toxicity data were standardized to account for the different laboratory conditions. These rankings were not significantly different for different heavy metals, allowing the aggregation of physiological sensitivity into a single ranking. Furthermore, the toxicological data for macroinvertebrates were used as input data to develop and validate prediction models for heavy metal toxicity, which are currently lacking for a wide array of species (Chapter 3). Apart from the toxicity data, the phylogenetic information of species (evolutionary relationships among species) and the physicochemical parameters for heavy metals were used. The constructed models had a good explanatory power for the acute sensitivity of species to heavy metals with the majority of the explained variance attributed to phylogeny. Therefore, the integration of evolutionary concepts (relatedness and similarity of species) with the chemical parameters used in ecotoxicology improved prediction models for species lacking experimental toxicity data. The ultimate goal of the prediction models developed in this thesis is to provide accurate predictions of toxicity for a wide range of species and chemicals, which is a crucial prerequisite for conducting chemical risk assessment. The latter was conducted for the first time on the continental scale (Chapter 4), by making use of a dataset of 4,000 sites distributed throughout 27 European countries and 91 respective river basins. Organic chemicals were likely to exert acute risks for one in seven sites analyzed, while chronic risk was prominent for almost half of the sites. The calculated risks are potentially underestimated by the limited number of chemicals that are routinely analyzed in monitoring programmes, and a series of other uncertainties related with the limit of quantification, the presence of mixtures, or the potential for sublethal effects not covered by direct toxicity. Furthermore, chemical risk was related to agricultural and urban areas in the upstream catchments. The analysis of ecological data indicated chemical impacts on the ecological status of the river systems; however, it is difficult to discriminate the effects of chemical pollution from other stressors that river systems are exposed to. To test the hypothesis of multiple stressors, and investigate the relative importance of organic toxicants, a dataset for German streams is used in chapter 5. In that study, the risk from abiotic (habitat degradation, organic chemicals, and nutrients enrichment) and biotic stressors (invasive species) was investigated. The results indicated that more than one stressor influenced almost all sites. Stream size and ecoregions influenced the distribution of risks, e.g., the risks for habitat degradation, organic chemicals and invasive species increased with the stream size; whereas organic chemicals and nutrients were more likely to influence lowland streams. In order to successfully mitigate the effects of pollutants in river systems, co-occurrence of stressors has to be considered. Overall, to successfully apply integrated water management strategies, a framework involving multiple environmental stressors on large spatial scales is necessary. Furthermore, to properly address the current research needs in ecotoxicology, a multidisciplinary approach is necessary which integrates fields such as, toxicology, ecology, chemistry and evolutionary biology.