Self-reported data on the municipality of residence were used to assess long-term exposure to outdoor air pollution from 1980 to 2002 in the longitudinal Canadian National Population Health Survey. Exposure to carbon monoxide, nitrogen dioxide, ozone, sulfur dioxide, and particulate matter was determined using data obtained from fixed-site air pollution monitors operated principally in urban areas. Four different methods of attributing pollution exposure were used based on residence in (1) 1980, (2) 1994, (3) 1980 and 1994, and (4) at all locations between 1980 and 2002. Between 1,693 and 4,274 of 10,515 members of the cohort could be assigned exposures to individual pollutants using these methods. On average, subjects spent 71.4% of the 1980-2002 period in the census subdivision where they lived in 1980. A single exposure measure in 1980 or 1994 or a mean of the two measures was highly correlated (r>0.7, P<0.0001) with a measure which accounted for all moves between 1980 and 2002. Although our ability to characterize long-term exposure was constrained by a lack of data from fixed-site monitors, the low frequency of moves meant that measures based on a single year generally provided a good approximation of long-term exposure at the census subdivision level.