In order to evaluate the benefit of arterialization of the portal vein in conjunction with a therapeutic end-to-side porto-caval shunt, we started in December 1979 a prospective randomized study, comparing these two techniques in Child class A and B cirrhotic patients with a hepatopetal portal flow of at least 100 ml per min. Sixty-four patients have been included in the study: 33 underwent a porto-caval shunt (NART) and 31 were arterialized (ART). The two groups of patients were similar as regards clinical and hemodynamic parameters. The operative mortality was lower after arterialization, considering the whole group (ART: 6.5%, NART: 12%) and high risk patients (Child class B: ART: 0%, NART: 22%--emergency operation: ART: 0%, NART: 17%) but the differences were not statistically significant. Postoperative ascites was more frequent in arterialized patients (ART: 45%, NART: 18%; p = 0.02), with an increased need for reoperation (ART: 26%, NART: 6%; p less than 0.05). The mean follow-up period is 56.9 +/- 28.1 months (SD) for the 58 surviving patients. The five-year actuarial survival rate is 68% for ART patients and 60.6% for NART patients (NS). In child B patients the five-year actuarial survival rate is 75% in ART patients and 22% in NART patients (p less than 0.05). Actuarial estimation of arterialization patency--proved by angioscan--is 38% at five years. There were no significant differences in the incidences of postoperative and long-term encephalopathy and liver function between the two groups. In conclusion, arterialization of the portal vein in conjunction with a therapeutic end-to-side porto-caval shunt improved survival in high risk patients (Child class B), did not increase operative mortality in Child A and B patients, was followed by a higher incidence of postoperative ascites and did not decrease the incidence of long-term encephalopathy.