BACKGROUND We studied the effect of intracoronary administration of arginine-8-vasopressin on blood flow in nondiseased coronary arteries and determined whether this vasoconstriction was severe enough to produce ischemia in 30 dogs. RESULTS In group 1 (n = 6), after vasopressin administration coronary blood flow was decreased by 41% (p less than 0.002) without changes in heart rate or aortic pressure, and left ventricular ejection fraction measured by radionuclide angiocardiography was decreased by 18% (p less than 0.0005). In group 2 (n = 6), ischemia was confirmed by measurement of transmural pH changes. Administration of vasopressin decreased subendocardial pH of the infused zone from 7.40 +/- 0.03 to 7.31 +/- 0.07 (p less than 0.01). The subendocardial pH of the zone not infused with vasopressin did not change. To overcome the intrinsic regulation of blood flow, operating primarily in small coronary arteries, we hypothesized that vasopressin must increase resistance primarily in large rather than small coronary arteries. After intracoronary infusion in group 3 (n = 6), however, most (94%) of the increase in resistance during vasopressin administration was explained by an increase of resistance in small coronary arteries. In group 4 (n = 9), vasopressin decreased coronary blood flow by 50% and decreased local shortening by 90% at a time when systemic hemodynamics were unchanged. Coronary constriction induced by vasopressin, or the recovery from it, also was not altered by cyclooxygenase blockade. CONCLUSIONS Thus, vasopressin produces myocardial ischemia by constricting small, nondiseased coronary arteries severely enough to overcome the competition from normal coronary regulation, and this ischemic event is not mediated by prostaglandin products.