Conducted as part of the Air and Health Monitoring Program (Psas), this study aims to analyze the short-term relationships between levels of four air pollution indicators (nitrogen dioxide—NO₂; ozone - O3, particulate matter - PM10 and PM2.5) and hospital admissions for respiratory and cardiovascular causes in eight French metropolitan areas (Bordeaux, Le Havre, Lille, Lyon, Marseille, Paris, Rouen, and Toulouse). Morbidity data were obtained by extracting records from the database of the Medical Information Systems Program (PMSI) of public hospitals, hospitals participating in the public health system, or private hospitals. Daily indicators of exposure to air pollution—NO2, O3, PM10, and PM2.5—were derived from concentrations measured in each study area by urban and suburban monitoring stations operated by accredited air quality monitoring associations. For each studied cause of hospital admission, risks were estimated by accounting for exposure on the day of the event and the preceding day (0–1-day exposure). For each exposure-risk relationship, a combined analysis of the locally obtained results allowed us to estimate a combined relative risk. We observed significant relationships between levels of particulate matter (PM10, PM2.5) and NO2 and the daily number of hospitalizations for cardiovascular causes. These associations are stronger for those aged 65 and older. They are also higher for cardiac causes, particularly ischemic heart disease, whereas they are not significant for cerebrovascular diseases. Regarding hospital admissions for respiratory causes, the excess relative risks associated with increased levels of NO2, PM10, and PM2.5 vary across the study areas. For these three pollution indicators, the combined excess risks across the 8 cities are positive but not significant. Ozone levels are significantly associated with the relative risk of hospitalization for respiratory causes only among people aged 65 and older. This study was conducted using a standardized methodology across the eight participating centers in terms of data collection, indicator construction, and statistical analysis. However, the PMSI, which remains primarily a medical-economic information system, represents the main limitation of this study. Indeed, the data it contains do not allow for distinguishing between scheduled admissions and emergency admissions. Furthermore, the failure of certain healthcare facilities to retain data beyond three years led to a shortening of the study period in some metropolitan areas. Thus, the lack of specificity of certain health indicators (particularly respiratory ones) and the small number of hospitalizations included may explain the wide confidence intervals surrounding the estimates of the observed relative excess risks. The results obtained here appear robust and valid with regard to hospitalizations for cardiovascular diseases. With regard to hospitalizations for respiratory causes, the lack of specificity of the health indicator used and the lack of statistical power in the analyses certainly explain the uncertainties surrounding the results obtained. In this context, the possibility of constructing more specific indicators (distinguishing admissions following a visit to the emergency department from other admissions) thanks to changes in the PMSI offers interesting prospects. Extending the retention period for PMSI data would also be an asset for this type of study. (R.A.)