We evaluated the effects of common vasoactive agents on collateral blood flow to an ischemic segment of small intestine and on the hemodynamic determinants of that flow. Two adjacent canine jejunal segments were isolated together, and the artery to each was cannulated for autoperfusion from a femoral and a carotid artery, respectively. Arterial pressure, arterial blood flow into, and venous outflow from each segment was measured separately. Venous pressure was zero. Vascular resistances were calculated. After clamping the arterial circuit to one segment, designated "ischemic," its steady-state venous outflow was taken as the collateral blood flow from the nonischemic into the ischemic segment. Without drugs, collateral blood flow was equal to 29 +/- 4 ml/min X 100 gm or, 56% +/- 8% of normal, well above the level needed to sustain oxygen consumption and thereby prevent ischemic injury. Local intra-arterial infusion of the vasodilators isoproterenol and papaverine not only failed to increase collateral flow but actually caused a small but (with isoproterenol) significant reduction, caused by vasodilation in the nonischemic bed, and a resulting drop in arterial pressure distal to the occlusion in the ischemic segment (i.e., a steal phenomenon). The vasoconstrictors levarterenol and vasopressin also reduced collateral flow but by direct and preferential vasoconstriction of the dilated ischemic bed. These findings suggest that collateral blood flow may be optimal without drugs and is decreased only by vasoactive agents, including vasodilators. This contradicts the rationale for vasodilator therapy for the direct augmentation of collateral blood flow in acute occlusive intestinal ischemia.