Myocardial perfusion imaging using positron emission tomography (PET) may be more accurate for the diagnosis of coronary artery disease (CAD) than conventional imaging. The purpose of this study was to evaluate the prognostic implications of perfusion abnormalities in 685 patients (age 62 +/- 11 years, 199 women) studied by PET, and to assess the incremental value of these data in relation to prognostic implications of clinical and angiographic findings. Rubidium (Rb)-82 PET was performed before and after dipyridamole stress. Transient defects were detected in 227 patients (33%), and were moderate or greater in severity (> 15% of the left ventricle) in 84 (12%). Resting defects were present in 435 (64%) and were moderate or greater in severity in 216 (32%). The total extent of abnormally perfused myocardium was small (< 15% of the left ventricle) in 198 (29%), moderate in 216 (32%), and extensive in 105 (15%). Clinic review or standardized phone interview in 657 patients (96%) identified 151 cardiac events, including 81 cardiac deaths, 16 patients with myocardial infarction, 7 with unstable angina, and 47 with late revascularization (> 3 months after PET). Normal scans had a 90% event-free survival, compared with 87% in patients with small, 75% with moderate, and 76% with extensive defects (log rank chi-square 30, p <0.0001). Functional class, extent of CAD, and the presence and extent of perfusion defects (both at rest and during stress) were independent predictors of cardiac death and total cardiac events. In sequential Cox proportional-hazards models, the results of PET were incremental to those of clinical and angiographic evaluation. Thus, the presence and extent of damaged and jeopardized myocardium are independent and incremental predictors of outcome in patients undergoing Rb-82 PET.