The relative impact of body mass index (BMI) and exercise capacity on mortality in patients with an intermediate to high likelihood of coronary artery disease (CAD) is not clear. Thus, the effect of BMI and exercise capacity on all-cause mortality in patients referred for stress myocardial perfusion imaging was investigated. The outcome of 2,119 patients undergoing exercise stress myocardial perfusion imaging from 1995 to 1999 was assessed. Patients lacked known CAD, but were at intermediate to high risk. Mortality outcome data were obtained from the Social Security Administration Death Master File. There were 183 deaths during an average follow-up of 8.4 +/- 1.4 years. A Cox proportional hazards model identified age, Bruce protocol exercise time, BMI, male gender, and diabetes mellitus as significant predictors of all-cause mortality. In multivariate analysis, both exercise capacity and BMI correlated inversely with mortality, with higher chi-squared impact related to exercise capacity than BMI. In conclusion, both increased exercise capacity and BMI were associated with lower mortality in patients with an intermediate to high likelihood of CAD after controlling for confounding variables, supporting an inverse impact of BMI on mortality. The origin for this "obesity paradox" is unclear.