The retrospective and prospective assessment of contaminant impacts at higher levels of biological organization (populations, communities) is today constrained by the difficulty to observe, detect and assess chemical effects directly at these integrated levels. This is because of both practical issues (time and spatial scale, extinction) and intrinsic properties of such complex biological systems (integration of multiple environmental influences, adaptability, stochasticity). To gain in sensitivity and specificity, the assessment of toxicity in the lab and in the field is thus performed considering effects on organisms (survival, fertility, growth, or sub-organism responses (biomarkers). The extrapolation from such lower tier toxicity tests to sensitivity of higher biological levels is therefore required to ensure the ecological relevance of toxicity assessment. In this framework, population modelling is proposed in order to bridge the gap between the alterations of individual performance traits and the potential impairment of population dynamics. In ecotoxicology, this modelling mechanistic demographic approach revealed theoretically and experimentally how toxic population impacts are deeply modulated by population life history. In this presentation based on examples from aquatic ecotoxicology, I illustrate how perturbation analysis of matrix population models allows to disclose key demographic transitions for population fitness and to recognize the interplay between these demographic sensitivities and toxicological sensitivities in the response of populations to chemical perturbations. I discuss how these modelling tools lead us to consider that between and within species differences in life history are key determinants of chemical effects in the ecosystems.