Associations between average total personal exposures to PM2.5, PM10, and NO2 and concomitant outdoor concentrations were assessed within the framework of the Genotox ER study. It was carried out in four French metropolitan areas (Grenoble, Paris, Rouen, and Strasbourg) with the participation, in each site, of 60-90 nonsmoking volunteers composed of two groups of equal size (adults and children) who carried the personal Harvard Chempass multipollutant sampler during 48 h along two different seasons ("hot" and "cold"). In each center, volunteers were selected so as to live (home and work/school) in three different urban sectors contrasted in terms of air pollution (one highly exposed to traffic emissions, one influenced by local industrial sources, and a background urban environment). In parallel to personal exposure measurements, a fixed ambient air monitoring station surveyed the same pollutants in each local sector. A linear regression model was accommodated where the dependent pollutant-specific variable was the difference, for each subject, between the average ambient air concentrations over 48 h and the personal exposure over the same period. The explanatory variables were the metropolitan areas, the three urban sectors, season, and age group. While average exposures to particles were underestimated by outdoor monitors, in almost all cities, seasons, and age groups, differences were lower for NO2 and, in general, in the other direction. Relationships between average total personal exposures and ambient air levels varied across metropolitan areas and local urban sectors. These results suggest that using ambient air concentrations to assess average exposure of populations, in epidemiological studies of long-term effects or in a risk assessment setting, calls for some caution. Comparison of personal exposures to PM or NO2 with ambient air levels is inherently disturbed by indoor sources and activities patterns. Discrepancies between measurement devices and local and regional sources of pollution may also strongly influence how the ambient air concentrations relate to population exposure. Much attention should be given to the selection of the most appropriate monitoring sites according to the study objectives.