W e have previously shown in anesthetized, open-chest dogs with coronary stenosis and endothelial injury that serotonin and/or thromboxane A2 (TXA2) receptor activation play a major role in the mediation of platelet-dependent, intermittent coronary occlusion. Using a similar model in awake, closed-chest dogs, we tested the following hypotheses: (a) treadmill exercise promotes the development of cyclic flow variations in dogs with coronary stenoses and endothelial injury; (b) ventricular pacing does not induce cyclic flow variations in the same dogs; and (c) TXA2 and/or serotonin are important mediators of exercise-induced cyclic flow variations in this model. The surgical preparation consisted of the application of a hard, flow-limiting constrictor and a Doppler ultrasonic flow probe around the left coronary artery of 11 dogs. Treadmill exercise resulted in the prompt development of cyclic flow variations in all 11 dogs. Ventricular pacing at rates as high as 170 beats/min induced cyclic flow variations in only one of five dogs. Exercise-induced cyclic flow variations were abolished by TXA2 and/or serotonin receptor antagonists in all but one dog. Thus, (a) treadmill exercise promotes the development of cyclic flow variations in dogs with coronary stenoses and endothelial injury; (b) ventricular pacing does not induce cyclic flow variations in most dogs in the same model; and (c) TXA2 and/or serotonin are important mediators of cyclic flow variations in this model.