Ultrasonography was used to evaluate the patency of portocaval shunts in 9 patients: 2 with spleno-renal anastomosis, 4 with portocaval truncular anastomosis, 2 with mesenteric-caval shunt and 1 with atypical shunt. The shunt count be demonstrated in 7 of these 9 patients. Angiography was performed in 6 cases and confirmed the results of ultrasonography--a finding favourable to the latter method and in agreement with those of other authors. However, the usefulness of ultrasonography varies according to the type of shunt created, as shown by our own experience and by published reports: 70% of portocaval truncular shunts, 65% of spleno-renal shunts and 46% of mesenteric-caval shunts are demonstrable by ultrasounds, these variations being due to differences in location and accessibility to the ultrasonic examination. It ultrasonography fails to display the anastomosis, it may provide indirect signs of patency. In addition to being harmless and relatively inexpensive, ultrasonography is a valuable method for evaluating portocaval shunts. It can easily be repeated and its results need not be confirmed by angiography.