The vast majority of patients who require liver transplants have significant portal venous hypertension, and their liver replacement is specifically aimed to treat three major complications of portal hypertension, namely, bleeding oesophageal varices, refractory ascites and encephalopathy (liver failure). Variceal haemorrhage and ascites can be treated effectively by various types of portal decompression surgery, but these procedures invariably decrease hepatic blood flow and hence worsen the function of an already compromised liver. The enthusiasm for treating oesophageal varices and ascites with non-selective portosystemic shunting started to wane more than two decades ago when it became apparent that the price of preventing haemorrhage from oesophageal varices and intractable ascites in this manner was dehumanizing encephalopathy and progressive hepatic dysfunction (Jackson et al, 1971; Resnick et al, 1974). With the development of potent and specific diuretic drugs, truly intractable ascites is extremely uncommon. When it does occur, more appropriate therapy than portosystemic shunt will often be a peritoneojugular shunt. Chronic hepatic encephalopathy can rarely be managed by medical means, and it is effectively treated only with liver transplantation. Thus, the principal problem in portal hypertension for which therapeutic planning must be done is the haemorrhage from oesophageal varices. Prophylactic and therapeutic uses of anti-P-adrenergic agents and endoscopic sclerotherapy play a significant role in the management of oesophageal varices. Various types of portosystemicvenous shunting have been evaluated as to their efficacy in preventing both haemorrhage from varices and accelerated decline of hepatic function. In our centre, where there is a chronic shortage of organs, Child’s A patients who have had variceal bleeding are systemically screened for selective splenorenal shunts, performed electively. The procedure is being used on a trial basis because it is attended by a lower incidence of encephalopathy than is seen after non-selective shunts. Forty patients have been entered with no perioperative deaths but longer follow-up will be necessary to assess the validity of this approach. Five of the 40 have come to successful transplantation during the first 12 months after their Warren