The 1999 health impact assessment in Brest and Nantes, related to chronic exposure to air pollution, is based on four steps: hazard identification, selection of exposure-risk relationships, exposure estimation, and risk characterization. The study characterizes: - the health benefit associated with a 25% reduction in pollution levels; - the expected impact of a 15% increase in pollution levels. The results are presented as the number of deaths attributable to air pollution. The health benefit is also expressed in additional years of life expectancy. The study in Brest is based on a single exposure-risk relationship. In Nantes, the air quality monitoring network is more comprehensive and allows for the use of four exposure-risk relationships. The health benefit associated with a 25% reduction in air pollution levels is interpreted as a conservative estimate of the health impact attributable to air pollution. This impact is estimated at approximately 38 (23–53) deaths in Brest and 40 (14–65) deaths in Nantes. It represents a reduction in life expectancy of 48 (29–68) days in Brest and 51 (17–84) days in Nantes. The uncertain assessment of exposures, the use of U.S. exposure-risk relationships, and strong assumptions to calculate the reduction in life expectancy mean that the results are subject to greater errors and uncertainties than in the conduct of short-term health impact assessments. (R.A.)