To evaluate the role of the use of different analytical techniques in determining differences in the reported effects of passive smoking in children and adolescents, data sets from 2 studies that had previously reported divergent results were subjected to a uniform analytic procedure using a first-order autoregressive model (AR1). Although the populations studied differed in age distribution and the frequency of both personal and maternal smoking, the assumptions of the AR1 model were met by both data sets, and the models developed fit both data sets well. Coefficients for personal smoking from the study-specific models were almost identical. An effect of maternal smoking on FEV1, which has been reported for the East Boston data, again was observed, but no such effect was observed for the Tucson data set. These findings suggest that the differences in reported results are not related simply to model specificity and that real differences in exposure or response to exposure may be occurring in different climatic situations. Such differences need to be taken into account in any assessment of indoor exposure to potentially noxious agents.