We report the strategy adopted for the modeling of a resident brown trout population living in a theoretical river network. We consider chronic cadmium pollution occurring in one stretch of river. Using RAMAS Metapop, spatial modeling allows us to integrate spatial heterogeneity influencing the population dynamics and also influenced by the heterogeneity of the environmental contamination. We show how to perform the assessment of global impact of pollution from local perturbations using this modeling step: we build for instance population level dose-response curves linking population features with pollutant concentrations. We take advantage of the flexibility of RAMAS Metapop to deal with demographic stochasticity in order to discuss the use of deterministic models or stochastic models (extinction risk concepts) to provide population endpoints for decision-making. Besides, we confirm the conclusion of a previous contribution focused on the potential effect of migratory disruptions due to pollution: random variations in the dispersal pattern during the breeding event expand the extinction risk of the trout population. So increasing efforts are necessary to develop knowledge relative to toxicant-induced spatial behaviors and to integrate such effects in the definition of environmental quality criteria.