Twenty patients with malignant ascites were treated by insertion of a peritoneovenous shunt. Twelve had a LeVeen shunt, 7 a Denver shunt and 1 patient had a LeVeen shunt subsequently replaced by a Denver shunt. Overall survival was poor, which reflects the advanced stage of the patients' disease. Function of the Denver shunt seemed to be better than that of the LeVeen shunt. None of the 8 Denver shunts inserted was a complete failure, while 4 out of 13 LeVeen shunts were of no benefit. With a satisfactorily functioning shunt, good palliation was always achieved. A Denver shunt has become our first choice.