Current techniques for the detection of recurrent coronary stenoses following bypass grafting have shown disappointing diagnostic accuracy. This study used the same dipyridamole-handgrip stress to compare the accuracy of rubidium-82 positron emission tomography and thallium-201 single photon emission computed tomography, in 50 consecutive post-bypass patients undergoing coronary arteriography at a mean interval of 6.5 years after surgery. Significant stenoses in native coronary vessels (greater than 50% diameter) or grafts (greater than 70% diameter) were defined by quantitative angiography. Forty-six patients had recurrent or residual stenoses, 43 (93%) had a perfusion defect identified by positron emission tomography, and 35 (76%) were identified by single photon emission computed tomography (P = 0.04). Fourteen of the 17 patients (82%) without previous Q-wave myocardial infarction were identified by positron emission tomography; 10 of the 17 (59%) were detected by single photon emission computed tomography (P = NS). Stress-induced perfusion defects were demonstrated by positron emission tomography in 19 patients; of this group, thallium imaging identified reversible defects in 11, showed no perfusion defect in 1, and portrayed a persistent defect in 7 patients. Significant graft disease was present in 33 patients; perfusion defects were identified by positron emission tomography in 30 (91%), and by single photon emission computed tomography in 24 (73%, P = NS). Four patients were fully revascularized, without significant recurrent coronary disease; normal perfusion was present in 3 (75%) by positron emission tomography, and 4 (100%) by single photon emission computed tomography.(ABSTRACT TRUNCATED AT 250 WORDS)