In this paper, the pooled NO2 association with nonaccidental mortality is examined across 10 cities in Canada in single- and two-pollutant time-series models. The results reaffirm that NO2 has the strongest association with mortality, particularly in the warm season. Although attributing such effects to NO2 cannot be ruled out, it is plausible that NO2 is acting as an indicator for some other exposure affecting the population. This could include PM2.5, as has been suggested from some personal exposure data, but it could also be indicating a more specific type of PM2.5, such as traffic-related particles, given that in cities the main source of NO2 is motor vehicle exhaust. NO2 could also be acting as a surrogate for other pollutant(s) originating from motor vehicles or high-temperature combustion, such as volatile organic compounds (VOCs) or polycyclic aromatic hydrocarbons. Another possibility is other oxidized nitrogen species ("NO(z)") or photochemically produced pollutants that can co-vary with NO2, especially during urban stagnation events. Data to test these different possibilities across several Canadian cities are examined. The focus is on correlations in time or space between NO2 and other pollutants that are more strongly linked to vehicle emissions. The results support the hypothesis that NO2 is a better indicator than PM2.5 of a range of other toxic pollutants. This includes VOCs, aldehydes, NO(z) and particle-bound organics in motor vehicle exhaust. Thus, overall, the strong effect of NO2 in Canadian cities could be a result of it being the best indicator, among the pollutants monitored, of fresh combustion (likely motor vehicles) as well as photochemically processed urban air.