Global left ventricular performance (ejection fraction) and regional function were studied by rest-exercise radionuclide ventriculography in 36 patients before and after (23 +/- 8.5 wk) they had coronary artery bypass surgery for stable angina pectoris. The exercise ejection fraction was less than the resting ejection fraction before surgery (n = 36,p = 0.006), but not after surgery. The degree of postoperative improvement correlated with the degree of preoperative dysfunction (r = 0.55, n = 36, p less than 0.001). Improvement was most likely to occur if exercise-induced dysfunction was present preoperatively (n = 15,p = 0.001), even with old myocardial infarction. Regional dysfunction during preoperative exercise was also likely to improve postoperatively (n = 18, p = 0.001). Protocol design is important in determining the results and their interpretation. Matching postoperative exercise loads to preoperative loads and using regional analysis with two imaging projections improved judgment of the results. Regional dysfunction was commoner than global dysfunction and was less sensitive to workloads than was ejection fraction. This study shows that coronary artery bypass surgery can improve left ventricular performance on exercise if preoperative tests indicate the presence of ischemia-induced dysfunction.