A cross-sectional study of 4,071 children aged 6-11 yr of age from a rural region of Western Pennsylvania was conducted in spring of 1979. Standardized children's questionnaires were distributed to the parents and returned by the children to school, where spirometry was performed. The region was divided into low-, moderate-, and high-pollution areas on the basis of the 1974-1978, 3-hr, 24-hr, and annual averages for sulfur dioxide (SO2). Seventeen monitoring stations in the region and a triangulation procedure were used to estimate centroid levels in each geographic residence area. After adjusting the respiratory symptom response outcomes and the pulmonary function levels for known predictors, no significant association was noted for level of SO2. However, the highest exposure categories were only slightly above the present annual and 24-hr National Air Quality Standards for SO2. We conclude that at levels of exposure to which these children were exposed, only by study of potentially sensitive subsets or measures of acute response would it be possible to detect respiratory outcomes associated with ambient air pollution.