Renal effects of the transjugular intrahepatic portosystemic shunt (TIPS) were compared in 6 patients without ascites (group 1), 11 patients with ascites responding to diuretic treatment (group 2), and 6 patients with refractory ascites (group 3). Seven days after insertion of TIPS, 24-hour urinary sodium excretion had increased in patients with ascites: 113 +/- 16 mmol to 170 +/- 30 mmol (P = .012) in group 2, and 22 +/- 8 mmol to 77 +/- 27 mmol (P = .050) in group 3. In group 3, fractional sodium excretion tended to increase from 0.26% +/- 0.14% to 0.62% +/- 18% (P = .081). The relative increase of urinary sodium excretion (to 444% +/- 122%) and fractional sodium excretion (to 413% +/- 127%) in group 3 was significantly (P < .05) higher than in group 1 and group 2, respectively. Creatinine clearance and 24-hour urinary volume were not significantly changed in either group. Patients with Child-Pugh class C had a more pronounced effect of TIPS on urinary sodium excretion (increase to 396% +/- 115% vs. 139% +/- 15%; P = .066) and on fractional sodium excretion (increase to 415% +/- 103% vs. 94% +/- 15%; P = .020) than patients with less-severe liver disease. Fractional sodium excretion of less than 0.35% before TIPS was found to be an indicator of renal response to TIPS. The effect of TIPS on urinary sodium excretion and on fractional sodium excretion was related to the patients' Child-Pugh score (r = .55; P = .007 and r = .68; P = .001, respectively) and inversely to their fractional sodium excretion (r = -.44; P = .047 and r = -.54; P = .012, respectively) before TIPS. These data demonstrate that TIPS affects renal sodium handling in patients with ascites, particularly in patients with refractory ascites. Severity of liver disease and fractional sodium excretion before TIPS are parameters to predict the extent of the renal response.