Our initial use of the distal splenorenal shunt (DSRS) in 1973 was fostered by disappointment with the results of so-called total shunts. This selective shunt was, when anatomically feasible, our preferred therapy until 1980, when surgical referral was affected by enthusiasm for sclerotherapy. Our study of 71 DSRSs is uncontrolled because we could not recruit patients for a prospective randomized trial that involved either no treatment of operations that had proven faults. Our experience shows that operative risk (4%) and incidence of postshunt encephalopathy (6%) are low, that the rate of shunt occlusion is acceptable (10%), and that bleeding is as well controlled as with other shunts. Survival rates correlate with the cause of portal hypertension and with hepatic functional reserve. Analysis of the causes of death shows that the natural history of cirrhosis and coexistent disease are major determinants of prognosis.