This paper employs the Behavioral Risk Factor Surveillance System (2001) data in conjunction with the Environmental Protection Agency's Air Quality System data to investigate how air pollution caused by motor vehicle emissions affects the likelihood of good health and the amount of health investments. Models are estimated using three different measures of overall health: a measure of self-assessed health and two health outcome indicators (asthma and blood pressure). A multivariate probit approach is used to estimate recursive systems of equations for self-assessed health, health outcomes and life-styles. The most interesting result concerns the influence of pollution on health-improving life-style choices: only if traffic pollution is in the ‘satisfactory range’ (AQI level at or below 100), individuals will have incentive to invest in health. Copyright © 2009 John Wiley & Sons, Ltd.