The Duke Treadmill Score (DTS) has been shown to predict mortality in women who have symptomatic heart disease, but its ability to do so in asymptomatic women is unknown, as is its comparative advantage to exercise capacity. We investigated whether a decreased DTS is associated with increased mortality in a prospective cohort of 5,636 asymptomatic women. A symptom-limited exercise treadmill test using Bruce's protocol was performed at baseline. DTS was calculated using exercise time, exercise-induced angina, and ST-segment depression. Exercise capacity was measured in METs. Deaths and cause of death were identified from 1992 to 2000. After adjusting for the Framingham Risk Score, the risk of death decreased by 9% for each unit increase in DTS and by 17% for every 1-MET increase (p <0.001). Those who had a DTS <5 (moderate or high risk) had hazard ratios for death and cardiac death that were 2.2 and 2.5 times greater, respectively, than did those who had a DTS > or =5 (low risk), after adjusting for Framingham Risk Score (p <0.001). Receiver-operating characteristic curves for the DTS model and the exercise capacity model were not significantly different. In conclusion, we have demonstrated that, although the DTS is an independent predictor of mortality and cardiac mortality in asymptomatic women, it does not appear to be a better predictor than exercise capacity alone. The role of ST-segment changes and symptoms with stress testing in asymptomatic women does not provide additional prognostic information.