Ecological risk assessment ought to make predictions about the likely effects of chemical contaminants on populations and ecosystems. However, current procedures either make these links on the basis of oversimplified approaches that obscure the relationship between what is measured in ecotoxicological tests and what we are trying to protect or are too demanding in terms of both data and understanding to be used practically and routinely. Here we develop an approach that uses simplified life-history models to make explicit and ecologically relevant links between test results and their implications for population dynamics and explore the possibility of extending this to an appreciation of ecosystem effects. These models also make a number of simplifying assumptions but in being explicit they are open to challenge and elaboration, and we argue that, in general, the assumptions are usually precautionary. The development of these models in itself gives some nonintuitive insights into the theory of ecotoxicological responses, and we illustrate the practical application of the theory using a number of hypothetical case studies.