The primary objective of the SAVIAH, a multi-centre study funded by European Union, was to assess new methodology for study of small area health statistics and to implement it in epidemiological health statistics and geography. In Prague, the study has been conducted in two city districts with large variation in air pollution. Data at individual level (health symptoms and socio-economic circumstances of the family) were collected by questionnaires completed by parents of 3680 children aged 7-10 both resident and attending schools within the area (response rate 88%). Aggregated data for geographical areas were available from census and urban planning sources for 692 enumeration districts in the study area which were aggregated into 75 medium sized areas. Outdoor concentrations of nitrogen dioxide (NO2) were monitored by passive samplers. All these data were integrated into a geographic information system (GIS). Spatial distribution of air pollution was estimated by kriging and multiple regression modelling. These models explained about 80% of the variation in air pollution measured by passive samplers. GIS was then used to assign to individuals an exposure based on place of residence and school in order to conduct individual based analyses. Association between NO2 and life-time prevalence of wheezing and/or whistling, and wheezing/whistling in the last 12 months was studied by logistic regression. For both outcomes, school levels of NO2 were positively related to symptoms but home levels of NO2 showed a negative association. Logistic regression at individual level gives similar results as ecological analysis and multilevel modelling. Hierarchical model yielded somewhat wider confidence limits. Adjustment for parental behavioural and socio-economic factors did not affect these estimates substantially. This study demonstrated the power of the GIS methodology in studying the effects of complex environmental factors on respiratory health of children.