OBJECTIVE: Our aim was to compare the results of gadolinium-enhanced breath-hold MR angiography with those of conventional angiography for the study of mesenteric circulation. SUBJECTS AND METHODS: MR angiography and digital subtraction angiography were prospectively performed in 33 patients referred for hepatic, pancreatic, or mesenteric disease. MR angiography was performed with four three-dimensional acquisitions at 0, 30, 60, and 90 sec after injection of 0.1 mmol/kg of gadolinium. Selective conventional angiography was used as the standard of reference. RESULTS: A pure arterial angiogram (one on which veins could not be visualized) was obtained in 27 patients during the second or third acquisition. By subtracting the arterial phase from an arteriovenous phase (third or fourth acquisition) we obtained a pure venous angiogram (one on which arteries could not be visualized) in 28 patients. Agreement was good or excellent for the hepatic artery (kappa = 0.78), the superior mesenteric artery (kappa = 0.65), the splenic artery (kappa = 0.70), the portal vein (kappa = 1.0), the superior mesenteric vein (kappa = 0.88), and the splenic vein (kappa = 0.75). Agreement was poor, and vessels were better shown by conventional angiography, for the intrahepatic arteries (kappa = 0.006) and the branches of the superior mesenteric artery (kappa = 0.14). MR angiography and conventional angiography revealed 29 and 27 portosystemic collaterals, respectively. CONCLUSION: Dynamic breath-hold contrast-enhanced MR angiography compared favorably with conventional angiography in preoperative assessment of the proximal mesenteric arteries and in the evaluation of portal hypertension; however, conventional angiography is still necessary to evaluate distal arteries.