In order to clarify debated issues of the medical treatment of ascites in cirrhosis--the usefulness of a low sodium diet and washout period preceding diuretic administration, maximal dosage of antimineralocorticoid to be reached before the addition of a loop diuretic, identifications of factors influencing treatment efficacy--115 hospitalized patients with non-azotemic cirrhosis and ascites were recruited and randomized to receive a diet providing either 40 or 120 mmol of sodium daily. After a washout period from the outpatient diuretic regimen for 7 days (Step 1), increasing dosages of K-canrenoate (200 mg/day every 4th day up to 600 mg) were administered to patients not undergoing spontaneous diuresis (Step 2). Upon the failure of Step 2, K-canrenoate (400 mg/day) and furosemide at increasing dosage (25-50-100 mg every other day) were given (Step 3). Nine percent of patients underwent spontaneous diuresis, and 77% developed a negative sodium balance by the end of Step 2 (69% with a dosage of K-canrenoate < or = 400 mg/day) and 93% by the end of Step 3. Two patients were withdrawn from the protocol due to diuretic side-effects. Univariate analysis showed that the type of diet did not influence the response to treatment. The washout period led to a significant increase in endogenous creatinine clearance; natremia significantly rose in hyponatremic patients. Multivariate analysis showed that creatinine clearance and plasma aldosterone were independent predictive factors of the response to treatment.(ABSTRACT TRUNCATED AT 250 WORDS)