Background/aims/methods-During hepatic vein catheterisation, in addition to measurement of hepatic venous pressure gradient (HVPG), iodine wedged retrograde portography can be easily obtained. However, it rarely allows correct visualisation of the portal vein. Recently, CO 2 has been suggested to allow better angiographic demonstration of the portal vein than iodine. In this study we investigated the eYcacy of CO 2 compared with iodinated contrast medium for portal vein imaging and its role in the evaluation of portal hypertension in a series of 100 patients undergoing hepatic vein catheterisation, 71 of whom had liver cirrhosis. Results-In the overall series, CO 2 venography was markedly superior to iodine, allowing correct visualisation of the different segments of the portal venous system. In addition, CO 2 , but not iodine, visualised portal-systemic collaterals in 34 patients. In cirrhosis, non-visualisation of the portal vein on CO 2 venography occurred in 11 cases; four had portal vein thrombosis and five had communications between diVerent hepatic veins. Among non-cirrhotics, lack of portal vein visualisation had a 90% sensitivity, 88% specificity, 94% negative predictive value, and 83% positive predictive value in the diagnosis of pre-sinusoidal portal hypertension. Conclusions-Visualisation of the venous portal system by CO 2 venography is markedly superior to iodine. The use of CO 2 wedged portography is a useful and safe complementary procedure during hepatic vein catheterisation which may help to detect portal thrombosis. Also, lack of demonstration of the portal vein in non-cirrhotic patients strongly suggests the presence of pre-sinusoidal portal hypertension. (Gut 2000;46:856-860)