To determine the predictive value of stress [13N]ammonia positron emission tomography (PET) for reversible ischemia, 31 patients with coronary artery disease underwent rest-stress [13N]ammonia PET before and after coronary artery bypass surgery. The circumferential profile analysis was applied to determine the presence of transient defect (TD) and persistent defect (PD) preoperatively, and the fate of perfusion abnormality and asynergy after the surgery was assessed. Preoperative PET demonstrated 100 segments with perfusion abnormalities, including 69 TD and 31 PD. Fifty-six of the 69 TD (81%) improved in regional perfusion, while only four of 31 PD (13%) improved in perfusion postoperatively (p less than 0.001). Of 75 segments showing regional asynergy on contrast or radionuclide ventriculography preoperatively, 34 of 48 segments with TD (71%) improved in asynergy, while only five of 27 segments with PD (19%) improved in asynergy postoperatively (p less than 0.001). Stress-delayed 201TI tomography was performed in 22 of them. The predictive values for improvement in perfusion (77%) and asynergy (65%) were similar as those by [13N]ammonia PET (81% and 71%, respectively). However, the predictive values for no improvement in perfusion and asynergy by 201TI tended to be low (66% and 58%, respectively), as compared to those in 13N ammonia PET (87%; p less than 0.05 and 81%; p = 0.09, respectively). We conclude that an accurate prediction of reversible ischemia and asynergy can be achieved with rest-stress [13N]ammonia PET. Particularly, it can identify irreversible areas more accurately than the commonly performed stress-delayed 201TI imaging.