The effects of discontinuing long-term diuretic therapy were investigated by means of a double-blind randomized controlled trial. A total of 141 elderly patients in the long-stay wards of six hospitals were found to be taking maintenance diuretics, and for 33 of these the drugs were judged to be mandatory. Of the remaining patients, 52 continued to receive diuretics while 54 were given matching placebo tablets. Eight in the latter group required diuretic therapy to be resumed during the following 12 weeks. The main change observed in patients who completed the trial was a slight increase of ankle oedema in the placebo group. Plasma potassium levels below 3.5 mEq/1 were found in some patients taking diuretics but not when the drugs had been withdrawn for 12 weeks. Blood pressure rose slightly in the placebo group and plasma urea rose slightly in the diuretic group. It was concluded that old people receiving long-term diuretic therapy without obvious current indication should have the drugs withdrawn under careful supervision so that those needing them could be identified.