OBJECTIVE The purpose was to determine the sensitivity, the specificity and a year-long risk of subsequent cardiovascular events in patients with low, intermediate and high risk prognostic Duke treadmill score (DTS) in comparison with the presence (or not) of the myocardial perfusion defects on radionuclide images. METHODS A prospective study, with 173 consecutive patients with 02 or more risk factors to coronary artery disease (CAD), who underwent to exercise single photon - emission computed tomographic myocardial perfusion images using technetium-99m tetrophosmin (SPECT) and treadmill test (with DTS), from one neighbourhood of Curitiba city, between January 2003 and February 2004, were followed up for cardiac-cause mortality and major cardiac events. Follow-up was performed in 13+/- 1 months and in 162 patients was complete. RESULTS The DTS mean those patients with cardiac event (18) over a year was -0.27 (95% CI= -3.97 to +3.91) and those free cardiac event patients (144) was +4.92 (95% CI= +4.03 to +5.81), with p<0.00069. The DTS sensitivity was 72.22% and the SPECT sensitivity was 77.78%, with no significant difference p=0.21. The DTS specificity was 54.17% and the SPECT specificity was 88.19%, with p<0.0001. The cumulative proportion free-events (Kaplan-Meier) curves demonstrated that 94% those patients with low-risk DTS remained free-cardiac events. In contrast, all high-risk DTS had adverse cardiac events. Those patients with intermediate-risk DTS had 15% of cardiac event over a year. CONCLUSIONS The DTS was as sensitivity as SPECT in determine a year risk for CAD. Those patients with DTS <-0.27 had high-risk cardiac event.