The explosion that occurred in Toulouse on September 21, 2001, at the AZF site is one of the most significant industrial accidents of recent decades, due to the force of the blast and its human and material consequences for the metropolitan area. In the days immediately following the incident, an epidemiological monitoring system to assess the health consequences of the explosion was established by the French Institute for Public Health Surveillance in collaboration with the Midi-Pyrénées Regional Directorate of Health and Social Affairs. The objectives of this epidemiological monitoring are: - to provide information to help adapt care for the population affected by this disaster and to formulate recommendations for the care of populations that may be exposed to similar events in the future. - to assess the medium- and long-term health consequences (beyond the initial tally of deaths and injuries in the first few days), in order to measure the extent of the lasting effects such an event may have on public health. Characterizing the health risks associated with the release of pollutants during and following the accident was one of the three main areas of focus for the initiative. Indeed, the explosion released a cloud of air pollution consisting primarily of nitrogen compounds that drifted over the southwestern part of the metropolitan area. Furthermore, the accident resulted in the release of nitrates into the branch of the Garonne River bordering the facility. Finally, the blast propelled particles and soil fragments from this former industrial site into nearby neighborhoods. The analysis of health risks associated with these releases therefore took into account various types of pollution in different environments simultaneously, over exposure periods of varying lengths depending on the contaminated environments. The effects were considered over different time scales (immediate and long-term effects), for various population groups (site workers, the general population, children), and across several geographic areas (near and far from the site). To provide the most comprehensive results possible and reduce uncertainties, the “health-environment” component of the epidemiological monitoring system simultaneously employed two methodological approaches: on the one hand, the approach of assessing health risks based on environmental, population-based, and literature data; and on the other hand, the specific surveillance of targeted conditions for early warning purposes through local health information systems. In the context of the major accident on September 21, this dual approach made it possible to characterize the risks associated with exposure to pollutants released during the explosion and in the weeks that followed. This report presents in detail the methods used in these two complementary approaches, the results obtained by each, and their interpretation in terms of health risk. In summary, the analysis does not suggest any major impact from exposure to pollutants released as a result of the accident. The estimated exposures in the various affected population groups and the health data obtained from the alert systems did not lead to recommendations for specific surveillance or precautionary measures other than those taken in the immediate aftermath of the disaster. This study also demonstrated the value of epidemiological intervention in an emergency context and the need to organize, in advance, the availability of essential information to address public health questions raised during such disasters. (R.A.)