Epidemiological studies have shown increased lung cancer risks in urban areas and in communities near some types of industries, such as non-ferrous smelters. The excess relative risks are generally of the order of 50% or lower, which implies that uncontrolled confounding has to be considered in the evaluation. Extrapolations of risks associated with occupational exposures provide some support for the findings. For cancer of other sites, the data are less consistent. An increased precision in exposure estimation is desirable in future epidemiological studies on cancer related to ambient air pollution.