Fifty-eight patients in established acute renal failure following trauma or surgery were allocated in a prospective and random fashion to two different diuretic regimes. In the control group, 1 g frusemide was given as a single injection over four hours. In the test group, frusemide was then continued either intravenously or orally in a dose of 3 g/24 hr until a urine output of 200 ml/hr was sustained or until the plasma creatinine fell below 300 mumoles/l. Oliguria was reversed or prevented in 24 of 28 patients given sustained frusemide, but in only 2 patients given a single injection. However, the number of dialyses and duration of renal failure and mortality were not different in the two groups. The serious complication of deafness occurred in two patients and in one of them this was permanent.