Abstract Peritoneovenous shunts have been inserted into 26 patients to control malignant ascites. All benefitted and most required no further paracentesis until death from progressive malignancy. Shunt blockage, which is the major problem at present, occurred in 8 patients. Five patients suffering from far advanced malignancy died within a month of operation. There was no clinical evidence of enhanced tumour spread or disseminated intravascular coagulation. We do not consider that the procedure is the first line of management, neither has it much to offer the patient with viscous, bloodstained or loculated ascites. We suggest criteria which help to identify the patient most likely to benefit from a peritoneovenous shunt.