There are few data sets appropriate for characterizing the indoor conc
entrations of air pollutants over the long tem. An understanding of th
e variability in indoor pollutant levels is particularly relevant to t
he design of epidemiologic investigations: misclassification of exposu
re due to the inaccuracy of exposure estimates tends to weaken the ass
ociation of exposure with health outcome. This paper uses a series of
indoor NO2 measurements collected at two-week intervals over 18-month
periods be 1988 and 1991 to describe the seasonal and year-to-year var
iability in indoor NO2. The data show that there can be large year-to-
year differences in both the sample distribution of indoor NO2 as well
as the household average. For homes with gas ranges with continuously
-burning pilot lights, the average bedroom NO2 concentration was 25% h
igher in the winter of 1990-1991 than in the winter of 1989-1990 but o
nly 4% higher during the winter of 1988-1989 than during the winter of
1989-1990. The winter-to-winter correlations within homes ranged from
a low of 0.53 to a high of 0.88. The year-to-year differences in mean
indoor concentrations were wt related to temperature. Occupant behavi
ors that influence air exchange rate and/or source use are hypothesize
d to be the major determinant of the observed pattern. Exposure data c
ollected during a single year should be cautiously extrapolated to oth
er years. However, in Albuquerque homes, the data suggest that the yea
r-to-year variability in household NO2 levels will not have a strong i
mpact on classifying exposure into broad categories.