Coronary blood flow and epicardial coronary artery diameter were simultaneously measured by an electromagnetic or Doppler flow probe and a pair of ultrasonic crystals, respectively, during reactive hyperemia in conscious dogs. Reactive dilation appeared after the full appearance of reactive hyperemia and lasted for a period of 4-20 times the duration of the coronary occlusion. beta-Receptor blockade (propranolol, 1 mg/kg iv) attenuated both the reactive hyperemia in volume by 21-22% (P less than 0.01) and dilative responses of the epicardial coronary diameter by 27-28% (P less than 0.01), despite a nonsignificant attenuation of the resting or peak hyperemic coronary blood flow. When coronary blood flow was held constant during reperfusion, by an occluder distal to the ultrasonic crystals, the reactive dilation disappeared. A peculiar reactive constriction was noted when coronary occlusion was performed proximal to the site of the ultrasonic crystals. Appearance of this constriction was at 149 and 385 s after the release of 5 and 60 s of coronary occlusion, respectively. This late reactive constriction disappeared after pretreatment with alpha- (phentolamine, 1 mg/kg iv) and/or alpha- + beta-blockade, but not with beta-blockade alone, and it was not observed when the coronary diameter was measured proximal to the occluder. Thus reactive dilation of the epicardial coronary artery derives from an increase in coronary flow and is reduced by propranolol via a reduction in the hyperemic flow, suggesting a flow-dependent change in the diameter of the epicardial coronary artery. Reactive constriction is a local phenomenon following marked reduction in the coronary diameter and is abolished by alpha-adrenergic blockade with phentolamine.