Abstract
This article provides an overview of recent U.S. cohort studies on air pollution and health and discusses design issues related to the study hypothesis, exposure assessment, confounder assessment, and effect modification. The article argues that, given the potential long-term effects of air pollution on survival and the enormous costs of achieving further air pollution reductions, it would be shortsighted not to invest in new cohort studies to address unresolved scientific questions. In addition, following the examples of Sweden, the Netherlands, and Germany, the use of geographical information systems is recommended to assess exposure to air pollutants at the address rather than the community level.