Background In previous unrandomized studies, we demonstrated that patients undergoing 8-mm diameter portacaval H-grafts with collateral ablation (partial shunts) have lower rates of portasystemic encephalopathy (PSE) postoperatively than patients undergoing total portacaval shunts. We postulated that nutrient hepatic blood flow was greater after partial shunts because 8-mm grafts preserved some portal flow. Methods To test this hypothesis, we analyzed hepatic hemodymunies in 18 of 30 randomized patients who consented to be studied after complete operative recovery, grouped according to shunt size. We measured nutrient hepatic blood flow using 99m-Tc-Mebrofenin tracer elimination kinetics, and fractionated it into its portal venous and hepatic arterial components. PSE was assessed by blinded observers. Stepwise logistic regression was used to select the variable that best predicted encephalopathy. From 1989 to 1993, we conducted a randomized, prospective trial of partial (8 mm) versus total (16 mm) portacaval H-grafts. Group differences were compared using one-way analysis of variance (ANOVA). Results Hepatic encephalopathy occurred in 2 of 10 patients with partial shunts versus 5 of 8 patients with total shunts. Nutrient hepatic blood flow was significantly higher for partial shunts compared with total shunts (403 ± 601 versus 243 ± 17mL/min). Three variables—nutrient hepatic blood flow, portal blood flow, and hepatic arterial flow—were analyzed by stepwise logistic regression. Nutrient hepatic blood flow was selected as the best predictor of hepatic encephalopathy. In this series, PSE did not occur in any patient with more than 325 mL/min of nutrient hepatic blood flow. Conclusion These findings provide a physiologic basis for clinical observations demonstrating a lower incidence of PSE with partial shunts and indicates the superiority of partial over total shunts. Partial shunts, by preserving portal flow, maintain higher nutrient hepatic blood flow than total shunts and thus minimize PSE rates.