Major air pollution accidents which occurred in the 1950s led to public awareness of the health hazards involved. Since that period, levels of air pollution have decreased, but several studies conducted in North America and Europe indicate that particulate air pollution is linked to increased cardiorespiratory morbidity and mortality. Despite this evidence, several questions were raised concerning the interpretation of the results (threshold effect, harvesting effect and biological plausibility). The aim of this review is to present the link between epidemiological findings and their use in health impact assessment. We review the main causal criteria applied to epidemiology in light of scientific evidence currently available. Some causality criteria are more important than others, but they all support the causal nature of the relationship between air pollution and health, and thus justify the feasibility of health impact assessment calculations. Recent studies on relative risk assessment show that even if the risk linked to worsening air quality is low, public health consequences are high. Such information must be made accessible to policy makers and the population in general so that, together with the public health workers, they can all contribute to improving air quality and health in their communities.