In 45 approximately 1-year-old purebred German shepherd dogs, the left circumflex and the right coronary arteries were chronically occluded by implantation of slowly swelling ameroid constrictors. Before the operation, 27 dogs were trained on a treadmill until they could run 8 mph on a 22% incline for 1 hour, 5 days per week. Two weeks after the operation, exercise was gradually resumed and continued for 4 weeks, until the preoperative fitness level had been regained. Preoperative exercise training lasted 1-3 months; postoperative training lasted 100 +/- 22 days (mean +/- SD). After the dogs had trained with two chronically occluded coronary arteries, collateral and coronary blood flows were measured with tracer microspheres at maximal coronary vasodilation (adenosine infusion) in an isolated, blood-perfused Langendorff preparation at perfusion pressures of 40, 60, 80, 100, 120 and 140 mm Hg. Eighteen nonexercising dogs that also had two-vessel coronary occlusion served as controls. Nine controls and nine exercising dogs were paired littermates. Exercise of relatively high intensity (heart rates greater than 200 beats/min) before and after occlusion had no effect on coronary collaterals. Collateral conductance in trained and untrained dogs reached only slightly less than 40% of that of the replaced coronary artery. This result agrees well with earlier results in a group of nonexercising dogs with chronic two-vessel occlusion studied in an identical way. Two-vessel occlusion was associated with a 25% mortality rate. All dogs died instantaneously without warning symptoms of ventricular fibrillation. Exercise had no influence on mortality.