1. In order to assess the prevalence and knowledge of diuretic prescription, 203 consecutive admissions to three general medical wards were interviewed. Additional information was collected on forms sent to general practitioners and hospital doctors responsible for each patient. 2. Prevalence of diuretic use was 31% (63 patients); other drugs only, 60% (121 patients); no drugs, 9% (19 patients). Patients gave an incorrect indication for 31% of diuretics prescribed, but for only 16% of other drugs (P < 0.005). 3. Compared with patients not taking diuretics, diuretic-users were older (mean +/- s.d., 70 years +/- 14 vs 54 +/- 20, P < 0.001), had lower abbreviated mental test scores (AMT scores seven or less, 21% vs 9%, chi 2 = 3.48, P = 0.06) and were prescribed more drugs (5.0 +/- 2.4 vs 3.2 +/- 2.0, P < 0.001). Increasing age and decreasing AMT score were associated with poorer knowledge of drug indication, but these factors could not explain fully the poor understanding of diuretic prescription. 4. Accurate recall of dose was the same for diuretics and for other agents. Knowledge about medication was not different if the general practitioner or hospital initiated treatment.