The impact of chronic urban air pollution on asthma and allergy remains controversial, probably because exposure is difficult to evaluate. In this cross-sectional epidemiological survey of 9- to 11-year-old children living in six French towns, levels of air pollution were estimated by using dispersion models that take into account background air pollution levels along with traffic density, topography and meteorological conditions. We estimated three-year exposure, in the school environment, to benzene, volatile organic compounds, SO2, particles less than 10 microns in diameter (PM10), nitrous oxides (NO2 and NOx) and CO, which are the main urban air pollutants. Several health indicators were gathered by means of a questionnaire, a physical examination, skin prick testing, and measurement of peak-flow variations following an exercise test. Among the nearly 5000 children who had been living in the survey area for at least three years, the 12-month and cumulative prevalence rates of asthma and exercise-induced hyper-reactivity correlated with estimated levels of benzene, SO2, PM10, NOx and CO. The 12-month and cumulative prevalence rates of atopic dermatitis correlated with estimated benzene, PM10, NO2, NOx and CO levels. Pollen sensitization correlated with benzene and PM10 levels, and the cumulative prevalence of allergic rhinitis correlated with the PM10 level. These associations remained statistically significant among children who had spent their entire lives in the same community.