BACKGROUND: The role of myocardial-perfusion imaging in calculating risk in symptom-free patients who have had coronary-artery-bypass grafting (CABG) is unclear. Practice guidelines have argued against routine screening of these patients. We sought to find out the independent and incremental prognostic value of exercise thallium-201 single-photon-emission computed tomography (SPECT) for prediction of death and non-fatal myocardial infarction (MI) in these patients. METHODS: Analyses were based on 873 symptom-free patients undergoing symptom-limited exercise thallium-201 SPECT between September, 1990, and December, 1993. All had undergone CABG and none had recurrent angina or other major intercurrent coronary events. Exercise and thallium-perfusion variables were analysed to determine their prognostic importance during 3 years of follow-up. FINDINGS: Myocardial-perfusion defects were noted in 508 (58%) patients. There were 57 deaths and 72 patients had major events (death or non-fatal MI). Patients with thallium-perfusion defects were more likely to die (9% vs 3%, p=0.0004) or suffer a major event (11% vs 4%, p=0.0002). Reversible defects were also predictive of death (12% vs 5%, p=0.002) and major events (13% vs 7%, p=0.004). The exercise variable with the strongest predictive power was an impaired (< or = 6 METs [measure of oxygen consumption equal to 3.5 mL/kg/min]) exercise capacity; poor exercise capacity was predictive of death (18% vs 4%, p<0.0001) and death or non-fatal MI (19% vs 5%, p<.00001). After adjusting for baseline clinical variables, surgical variables, time elapsed since CABG, and standard cardiovascular risk factors, thallium-perfusion defects remained predictive of death (adjusted relative risk 2.78, 95% CI 1.44-5.39) and major events (2.63, 1.49-4.66). Similarly, impaired exercise remained strongly predictive of death (4.16, 2.38-7.29) and major events (3.61, 2.22-5.87) after adjusting for confounders. INTERPRETATION: In this group of patients who were symptom-free after CABG, thallium-perfusion defects and impaired exercise capacity were strong and independent predictors of subsequent death or non-fatal MI. Recommendations against routine screening exercise myocardial-perfusion studies in this setting should be reconsidered.