Since 1971, 70 patients have been seen at Emory University Hospital with gastroesophageal varices secondary to extrahepatic portal vein thrombosis (PVT). Thirty-seven of these patients had had prior major operative therapy. In only three patients (8%) was shunt surgery successful, and there was a high incidence of rebleeding, other morbidity, and mortality. Of especial note are the serious consequences of simple splenectomy; splenomegaly and thrombycytopenia should rarely, if ever, be used as indication for splenectomy in portal hypertension. In 1977, the use of selective distal splenorenal shunt (DSRS) was begun at Emory in this population and a selective shunt has been possible in 24 of 29 patients (83%) who had had no prior operative therapy. Results have been excellent with a greater than 90% patency rate, long-term portal perfusion in all, no encephalopathy, and late rebleeding in one patient. Quantitative studies at 3-6 years show stability of liver function, significant decrease in spleen size, and rise in platelet count. However, long-term follow-up (greater than 15 years) is required in PVT patients before definitive assessment can be obtained. A specific problem of the PVT patient is late shunt stenosis which requires close observation; dilatation of the shunt was performed in six of the 24 patients with a patent shunt. Poor results with non-shunt operative procedures in PVT were again documented. The proper role of endoscopic variceal sclerotherapy is not yet clear, but appears to be an excellent addition to the therapeutic options. In conclusion, for patients with a patent splenic vein, initial therapy should be a selective shunt; for patients without a patent splenic venous system, endoscopic sclerotherapy is the procedure of choice.