We studied different algorithms to identify patients with heart failure who could potentially benefit from revascularization. Thirty-five coronary artery bypass (graft) patients with an ejection fraction of 35 ± 7% underwent preoperative <sup>18</sup>F-fluoro-2-deoxyglucose positron emission tomography (PET), low-dose dobutamine echocardiography (LDDE), and exercise testing. Follow-up by echocardiography and coronary angiography was performed 6 months after coronary artery bypass grafting. The sensitivity for prediction of reversible myocardial dysfunction was highest for PET and for ST depression or angina pectoris during exercise testing (100 and 93%, p = NS), 71% for LDDE (p < 0.05 vs. PET), and 50% for resting ECG (p < 0.02 vs. PET and exercise test). The specificity did not differ between LDDE (81%), PET (67%), and resting ECG (71%), but was lowest for exercise testing (33%; p < 0.02 vs. PET, LDDE, and resting ECG). Accuracies were: PET 80%, LDDE 77%, exercise testing 62%, and resting ECG 58% (p < 0.05 vs. PET). In patients with a negative exercise test, recovery was unlikely, and further viability testing may not be needed. In patients with a positive test, recovery may occur, and additional PET or LDDE should be performed. In these cases, PET with an <sup>18</sup>F-fluoro-deoxyglucose uptake of ≧70% as the criterion for viability yields optimum diagnostic characteristics. This strategy awaits further evaluation in larger patient populations with heart failure.