INTRODUCTION: Annoyance caused by air pollution has been proposed as an indicator of exposure to air pollution. The aim of this study was to assess the geographical homogeneity of the relationship between annoyance and modelled home-based nitrogen dioxide (NO2) measurements. METHODS: The European Community Respiratory Health Survey II was conducted in 2000-1, in 25 European centres in 12 countries. This analysis included 4753 subjects (from 37 in Tartu, Estonia, to 532 in Antwerp, Belgium) who answered the annoyance question and with available outdoor residential NO2 (4399 modelled and 354 measured) including 20 cities from 10 countries. Annoyance as a result of air pollution was self-reported on an 11-point scale (0, no disturbance at all; 10, intolerable disturbance). Demographic and socioeconomic factors, smoking status and the presence of respiratory symptoms or disease were measured through a standard questionnaire. Negative binomial regression was used. RESULTS: The median NO2 concentration was 27 microg.m(-3) (from 10 in Ume?Sweden, to 57 in Barcelona, Spain). The mean of annoyance was 2.5 (from 0.7 in Reykjavik, Iceland, to 4.4 in Huelva, Spain). NO2 was associated with annoyance (ratio of the mean score 1.26 per 10 microg.m(-3), 95% CI 1.19 to 1.34). The association between NO2 and annoyance was heterogeneous among cities (p for heterogeneity <0.001). CONCLUSIONS: Annoyance is associated with home outdoor air pollution but with a different strength by city. This indicates that annoyance is not a valid surrogate for air pollution exposure. Nevertheless, it may be a useful measure of perceived ambient air quality and could be considered a complementary tool for health surveillance.