Avian influenza is a contagious infectious disease caused by viruses of the family Orthomyxoviridae, influenza A viruses, affecting birds and known to cause extremely high mortality, especially in poultry. Highly pathogenic strains of avian influenza viruses cause severe and fatal disease in poultry and pose a risk to public health. In February 2006, France was hit by a highly pathogenic H5N1 avian influenza outbreak, affecting aquatic wildfowl in Dombes wetland. The Dombes is a wetland mosaic of more than a thousand ponds and an area of international ornithological importance, home to thousands of waterfowl. This is an area at risk for the spread of avian influenza viruses due to both the presence of wild birds on the ponds and the proximity of these populations to poultry farms that are present in the area. The main objective of this work is the modeling of the avian influenza risk in the Dombes region. This work is focused on the wildfowl and the risk of infection and transmission of infection by wild birds. As a first step, we have identified the specific characteristics of the Dombes area, which may affect the epidemiology of avian influenza in the region. The three main players in the Dombes ecosystem are the ponds, the wildfowl populations and the poultry farms. Secondly, the avian influenza risk in the Dombes was declined at two levels: 1) in the ecosystem of a single pond; and 2) within a network of ponds as characterized by 1), where the infection is spread from pond to pond. At the pond level, the avian influenza risk is modeled, on the one hand, by the persistence time of avian influenza viruses in the water of the pond, and, on the other hand, by the infection dynamics within wild bird populations by direct transmission between birds and indirectly through the water of the pond. At the ponds network, the risk analysis identifies spatial clusters of ponds at high risk of infection with avian influenza viruses. Concerning the risk of exposure of poultry farms to avian influenza viruses of wildfowl origin, it is calculated taking into account both the proximity of the farms to ponds likely to be infected and the probability of these ponds to become infected during an outbreak.