Anthropogenic stressors, such as pollution and habitat degradation, have affected natural fish populations (and other species) by increased mortality and/or impaired reproduction. Depending on the life history of a population, the sensitivity to a particular stress varies. Demographic population models can be used to evaluate these differences and determine which populations are most sensitive to a specific impact. In the present study, five populations of roach (Rutilus rutilus) with different life histories were modeled and compared for sensitivities to reductions in fecundity or survival. The five populations reflected a north-south gradient with increasing life span, delayed maturity, and reduced fecundity at higher latitudes. Furthermore, three types of population models were used and compared. The population models were an unstructured model, a two-stage model with juveniles and adults and an age-structured model with age-dependent fecundity. The results of the population models showed that populations at higher latitudes are more sensitive to a reduction in survival but less sensitive to a reduction in fecundity. Current methods in ecological risk assessment are based on oversimplifications due to a lack of ecological theory. The use of population models may increase the ecological relevance and help produce better risk assessments. The two-stage model provided results that were similar to the results of the more complex agestructured model in most cases. The unstructured model, however, deviated significantly from the other two models.