OBJECTIVE To investigate the potential correlation between ambient air pollution exposure and emergency department (ED) visits for depression. METHODS A hierarchical clusters design was used to study 27 047 ED visits for depression in six cities in Canada. The data used in the analysis contain the dates of visits, daily numbers of diagnosed visits, and daily mean concentrations of air pollutants as well as the meteorological factors. The generalized linear mixed models technique was applied to data analysis. Poisson models were fitted to the clustered counts of ED visits with a single air pollutant, temperature and relative humidity. RESULTS Statistically significant positive correlations were observed between the number of ED visits for depression and the air concentrations of carbon monoxide (CO), nitrogen dioxide (NO2), sulphur dioxide (SO2) and particulate matter (PM10). The percentage increase in daily ED visits was 15.5% (95% CI: 8.0-23.5) for CO per 0.8 ppm and 20.0% (95% CI: 13.3-27.2) for NO2 per 20.1 ppb, for same day exposure in the warm weather period (April-September). For PM10, the largest increase, 7.2% (95% CI: 3.0-11.6) per 19.4 ug/m3, was observed for the cold weather period (October-March). CONCLUSIONS The results support the hypothesis that ED visits for depressive disorder correlate with ambient air pollution, and that a large majority of this pollution results from combustion of fossil fuels (e.g. in motor vehicles).