OBJECTIVES About 20% of elderly people use long-term diuretic medication, but there is doubt whether prolonged diuretic medication on such a large scale is necessary. We performed a study to assess what proportion may successfully be withdrawn from diuretic therapy. DESIGN Double blind randomised controlled trial with six month follow up. SETTING General practice. SUBJECTS 202 patients taking long-term diuretics without manifest heart failure or hypertension. INTERVENTIONS Patients were allocated to either placebo (withdrawal group, n = 102) or continuation of diuretic treatment (control group, n = 100). MAIN OUTCOME MEASURE Occurrence of clinical conditions requiring diuretic therapy based on fixed criteria. RESULTS During follow up diuretic therapy was required in 50 patients in the withdrawal group and 13 in the control group (risk difference 36%; 95% confidence interval 22% to 50%). Heart failure was the most frequent cause of prescribing diuretic therapy (n = 25). Cessation of diuretic therapy caused a mean increase in systolic blood pressure of 13.5 (9.2 to 17.8) mm Hg and in diastolic pressure of 4.6 (1.9 to 7.3) mm Hg. CONCLUSION Withdrawal of long-term diuretic treatment in elderly patients leads to symptoms of heart failure or increase in blood pressure to hypertensive values in most cases. Any attempt to withdraw diuretic therapy requires careful monitoring conditions, notably during the initial four weeks.