Partial portosystemic shunts have been popularized because of a reported low rate of mortality and morbidity (especially encephalopathy, liver failure and occlusion). The results of partial portacaval shunts [small-diameter expanded polytetrafluoroethylene (ePTFE) H-graft portacaval shunt] were retrospectively reviewed to evaluate the clinical efficacy in the treatment of portal hypertension. Forty-three patients with portal hypertension were treated by small-diameter H-graft of ePTFE portacaval shunt from May 1995 to April 2006. Thirty-three had externally ringed grafts and ten had non-ringed ones. Ten had grafts of 10 mm in diameter and 33 had grafts of 8 mm. The left gastric artery and coronary vein were ligated in all the cases. Six had pericardial devascularization and splenectomy was performed in 42. An average decrease of free portal pressure (FPP) from (33.24 ± 4.78) cmH2O before shunting and (13.65 ± 5.65) cmH2O after shunting was observed. The portal blood flow was reduced by one-third of that before shunt. Thirty-eight patients survived and no upper gastro-intestinal rebleeding occurred in the follow-up period (50.5 months in average). Two were out of contact. Color Doppler ultrasonography and/or portography revealed the shunts were patent in 38 cases and were occluded in three cases (3/41, 7.3%). Encephalopathy developed in five cases (5/41, 12.2%). Partial (small-diameter ePTFE H-graft) portacaval shunting can reduce the portal pressure effectively. Majority of the hepatic flow from the portal vein can be maintained adequately. The shunts with reinforced grafts can keep a higher rate of patency. The morbidity of encephalopathy was lower than those with total shunt. The partial portacaval shunt is effective in preventing recurrent variceal bleeding.