It has recently been suggested that insertion of a titanium tip at the venous end of the LeVeen shunt drastically reduces the rate of shunt obstruction. To assess whether the LeVeen shunt with titanium tip improves the results obtained with therapeutic paracentesis, 81 patients with cirrhosis and refractory ascites were randomly assigned to therapy with paracentesis plus intravenous albumin (42 patients) or LeVeen shunt with titanium tip (39 patients). If patients were readmitted for ascites during follow-up, those in the first group were treated with paracentesis, and those in the LeVeen shunt group by the insertion of a new valve or a new shunt if obstruction was demonstrated. During first hospitalization, both treatments were equally effective in removing ascites. Complications were similar in both groups except for a higher rate of severe bacterial infection in the LeVeen shunt group. The mean duration of hospitalization was shorter in the paracentesis group than in the shunt group. During follow-up, the total number of readmissions and the number of readmissions for ascites were higher in the paracentesis group than in the shunt group (252 vs. 99, P < .001; and 193 vs. 43, P < .001, respectively). The total time in hospital, however, was similar (38 +/- 38 vs. 39 +/- 43 days, P = NS). Three patients had obstruction of the shunt during first hospitalization and 14 patients had a total of 22 obstructions during follow-up. Long-term survival was similar in both groups. The insertion of a titanium tip at the venous end of the LeVeen shunt does not prevent obstruction.(ABSTRACT TRUNCATED AT 250 WORDS)