1. A new isolated perfused preparation is described that allows a direct comparison to be made of the responses of the perfused arterial and retrogradely perfused venous circulations of the rat superior mesenteric vascular bed. 2. In experiments comparing the responses of the intact arterially perfused mesentery and small intestine to those of the same preparation following removal of the intestine and division of the circulations, the increases in perfusion pressure produced by arginine-vasopressin (30 pmol) and noradrenaline (1 nmol) were retained by the arterial circulation and those induced by angiotensin II (30 pmol) by the venous circulation. Endothelin-1 (30 pmol) constricted both portions of the vasculature but the prolonged nature of its response was associated with only the venous vessels. 3. In the simultaneously perfused arterial and venous preparation arginine vasopressin (3-100 pmol) was a selective constrictor of the arterial circulation and angiotensin II (3-100 pmol) of the venous circulation. In addition, noradrenaline (0.3-10 nmol), 5-hydroxytryptamine (0.3-10 nmol) and KCl (1-60 micromol) were more active as constrictors of the arterial than the venous vessels, and U46619 (10-300 pmol) a more active constrictor of the venous than the arterial vessels. Endothelin-1 (3-100 pmol) constricted both the arterial and venous portions of the vasculature but was significantly longer acting as a venoconstrictor than an arterioconstrictor. 4. Angiotensin I (300 pmol) caused constrictions of the venous circulation which were dependent upon the presence of angiotensin converting enzyme for captopril (10 microM) abolished constrictions caused by angiotensin I but not by angiotensin II. 5. In preparations preconstricted by U46619 (0.3-3 microM), acetylcholine (0.01-100 nmol), bradykinin (0.001-nmol), sodium nitroprusside (0.01-lOnmol) or isoprenaline (1-l00pmol) produced dose-related dilatations of both the arterial and the venous vasculatures, whereas adenosine diphosphate (ADP, 0.01-lOOnmol) caused dose-dependent dilatations of the arterial circulation but principally constrictions of the venous circulation. The dilatations caused by acetylcholine and bradykinin in both portions of the circulation, and by ADP in the arterial circulation, were endothelium-dependent as they were inhibited by gossypol (3 microM), whereas dilatations to sodium nitroprusside were not. 6. This preparation allows the responses of the arteries and veins of a single perfused mesenteric bed to be compared. In addition, with this preparation it is possible to demonstrate that veins, as well as arteries, show significant endothelium-dependent relaxations. It is concluded that the venous portion of the vasculature is significantly involved in the responses of the intact circulation.