Elevated lung cancer mortality rates in Harris County, Texas compared with other US counties and previously published reports that suggested a causal relation between air pollution and lung cancer in Houston prompted this ecologic analysis. A weighted regression analysis was used to examine the air pollution-lung cancer mortality relation for white males in Harris County, Texas, 1979-1981. The regression model included the following census tract-specific characteristics: median age for white males, two social and demographic factors not strongly correlated with pollution (family life cycle and migration), an age-dependent smoking index, and a pollution measure based on total suspended particulates. This model indicated a statistically significant contribution of the pollution measure in explaining the intracounty variation in lung cancer mortality rates. The relation between air pollution and lung cancer mortality, however, appeared to be highly dependent on which social and demographic factors were selected for inclusion in the analysis. Air pollution was not demonstrated to be a strong determinant of lung cancer mortality in this study in that the presence of air pollution accounted for less than 5% of the total variation in intraurban lung cancer mortality. In addition, the interpretation of geographic analysis must be guarded due to the introduction of potential bias due to aggregation. The hypothesis that air pollution is contributory to lung cancer cannot be tested until other, stronger individual risk factors for lung cancer can be better measured and controlled in studies of this association.