Most health studies measuring indoor nitrogen dioxide (NO2) concentrations have utilised long-term passive monitors. However, this method may not provide adequate information on short-term peaks, which may be important when examining health effects of this pollutant. The aims of this study were to investigate the relationship between short-term peak (peak) and long-term average (average) NO2 concentrations in kitchens and the effect of gas cookers on this relationship. Both peak and average NO2 levels were measured simultaneously in the kitchens of 53 homes using passive sampling techniques. All homes were non-smoking and sampling was conducted in the summer months. Geometric mean (95% confidence interval (CI)) average NO2 concentrations for all homes were 16.2 mu g m(-3) (12.7-20.6 mu g m(-3)). There was no difference between homes with and without gas cookers (p = 0.40). Geometric mean (95% CI) peak NO2 concentrations were 45.3 mu g m(-3) (36.0-57.1 mu g m(-3)). Unlike average concentrations, peak concentrations were significantly higher in homes with gas cookers (64.0 mu g m(-3), 48.5-82.0 mu g m(-3)) compared to non-gas homes (25.1 mu g m(-3), 18.3-35.5 mu g m(-3)) (p < 0.001). There was only a moderate correlation between the peak and average concentrations measured in all homes (r = 0.39, p = 0.004). However, when the data were analysed separately based on the presence of gas cookers, the correlation between peak and average NO2 concentrations was improved in non-gas homes (r = 0.59, p = 0.005) but was not significant in homes with gas cookers (r = 0.19, p = 0.33). These results suggest that average NO2 concentrations do not adequately identify exposure to short-term peaks of NO2 that may be caused by gas cookers. The lack of peak exposure data in many epidemiological studies may explain some of the inconsistent findings. (c) 2006 Elsevier Ltd. All rights reserved.