Twenty patients with cirrhosis and ascites but no renal failure were given piretanide, a new loop diuretic, in order to investigate its efficacy and to relate the diuretic response with the pretreatment plasma aldosterone concentration. Eleven patients responded to piretanide 12 mg/day (equivalent in potency to 80 mg furosemide); there was no response in nine patients. Both groups were similar with regard to liver function, plasma urea, serum creatinine, plasma electrolytes, urine volume, and urine potassium concentration. The basal urinary sodium excretion was significantly higher in those patients who responded (23.6 +/- 5.7 mmol/day vs. 4.3 +/- 1.42 mmol/day; P < 0.01) (M +/- SE). Plasma renin activity (PRA) and plasma aldosterone concentration (PAC) were normal or only slightly increased in patients who responded to piretanide (PRA = 1.22 +/- 0.20 ng/ml/h; PAC = 12.25 +/- 2.20 ng/100 ml) and very high in patients who did not respond (PRA = 8.71 +/- 1.18 ng/ml/h; PAC = 84.6 +/- 16.2 ng/100 ml) (P < 0.001). Patients unresponsive to piretanide 12 mg/day also failed to respond when the dose was increased to 24 mg/day. However, the addition of spironolactone, 150 mg/day, to piretanide was followed in these patients by a marked increase in diuresis and natriuresis. These results strongly suggest that the pre-treatment level of aldosterone is an important factor influencing the response to loop diuretics in patients with non-azotaemic cirrhosis and ascites.