The aims of distal splenorenal shunt with splenopancreatic disconnection (DSRS-SPD) were to improve maintenance of portal flow and prevent siphoning of hepatotrophic factors from the pancreas, as occurs after standard DSRS. The main patient population targeted for improvement were alcoholic cirrhotics, who have poorer survival than nonalcoholic cirrhotics and greater loss of portal flow (60%) after standard DSRS. Seventy-eight patients had DSRS-SPD during the study period 1983 to 1987: thirty-two patients were Child's A, 25 were Child's B, and 21 were Child's C. The 35 patients with alcoholic cirrhosis were a significantly poorer risk group by Child's class and galactose elimination capacity (GEC) than the 39 patients with nonalcoholic cirrhosis. Four patients had portal vein thrombosis. At 4-year follow-up, portal perfusion is maintained in 84% alcoholic and 90% nonalcoholic patients, with hepatic and systemic hemodynamics showing identical patterns for both groups. Hepatic function measured by GEC was maintained in alcoholic patients (290 +/- 68 mg/min to 303 +/- 74 mg/min) and nonalcoholics patients (342 +/- 92 to 320 +/- 118 mg/min). Gastric variceal rebleeding occurred in 10 patients--4 early (less than 2 months) and 6 late (18 to 54 months), leading to operation in 4 and transhepatic embolization in 4 patients: 2 of these patients died from this complication. Survival data show an operative mortality rate of 6.4% and overall mortality rate of 30%, with no significant difference between alcoholic and nonalcoholic cirrhotics. DSRS-SPD has significantly improved maintenance of portal perfusion and survival in patients with alcoholic cirrhosis requiring selective shunt for variceal bleeding when compared to standard DSRS. In this population DSRS-SPD is the operation of choice. In patients with nonalcoholic cirrhosis, the current data have not shown DSRS-SPD to have advantage over standard DSRS.