To determine the physiological effect of coronary artery bypass surgery and the mechanisms for pain relief, 15 patients with exertional angina were studied before and after operation. Before the operation conventional tests included exercise tests (all positive) and coronary angiography (all patients had greater than or equal to 70% stenosis of major vessels). In addition, ambulatory electrocardiographic monitoring during 48 hours detected 92 episodes (greater than or equal to 1 mm) of ST depression. Regional myocardial perfusion was assessed with positron tomography using rubidium-82 (t1/2 78 s) and this showed reversible inhomogeneity with absolute regional reduction of cation uptake after exercise in all 15 patients. After coronary surgery 10 of the 15 patients had (a) no angina, (b) patent grafts (three or more), (c) no evidence of ischaemia during ambulatory monitoring out of hospital, and (d) homogeneous perfusion with reversal of the disturbances in regional myocardial perfusion after exercise. After operation one of the 15 patients had no angina and showed silent infarction in the segment that was previously ischaemic but supplied by a patent graft. All but one of the remaining patients had no angina, patent grafts, but disturbances of regional myocardial perfusion with silent ischaemia on exercise. Two of these patients continued to have asymptomatic and ischaemic episodes of ST depression during ambulatory monitoring out of hospital. This physiological study of regional myocardial perfusion in patients in hospital and in those with ischaemia out of hospital showed that three different mechanisms may account for the relief of pain--improved perfusion, infarction, and silent ischaemia. Silent ischaemia in particular raises puzzling pathophysiological and therapeutic questions that may affect prognosis and the interpretation of clinical trials.