Leukopenia, thrombocytopenia, and hemolytic anemia occur commonly in advanced cirrhosis. Some investigators have reported that portacaval anastomosis (PCA) abolished hypersplenism while others have not found PCA to be uniformly beneficial. We compared the frequency of hypersplenism before and after admission to a controlled investigation of the effects of PCA in 52 unoperated control subjects and 38 patients with patent PCA. The two groups were followed for an average period of 5 1/2 years. On admission to the study leukopenia was present in about 2% of patients, thrombocytopenia in 6%, and hemolytic anemia in 4%. Splenomegaly was present in 48% and hypersplenism in 11%. After randomization splenomegaly disappeared more frequently in the shunted group. In addition, fewer patients with PCA developed splenomegaly for the first time after inclusion into the study than did unoperated control subjects. Leukopenia, thrombocytopenia, and hemolytic anemia, when present at inclusion into the study, disappeared with equal frequency in the shunted and unshunted patients, and appeared with equal frequency in both groups after randomization in previously unaffected patients. In no instance was hypersplenism clinically significant nor was splenectomy considered or carried out in any of these 90 patients. In additional uncontrolled studies we observed that therapeutic PCA did not affect hypersplenism differently from prophylactic PCA. We conclude that PCA has neither clinically nor statistically significant effects on hypersplenism.