A placebo-controlled study was employed to evaluate the effects of oral amrinone in patients with congestive heart failure. After a baseline period of at least 4 weeks of standard treatment for refractory congestive heart failure, oral amrinone was added to the treatment regimen of 173 patients. Patients were predominantly male (89%), aged 24 to 76 years (mean 54), with ischemic (52%) or idiopathic (37%) dilated cardiomyopathy, in New York Heart Association functional class II (40%), III (59%) and IV (1%) and having a mean (+/- standard deviation) left ventricular ejection fraction of 25 +/- 15%. Phase 1: After the addition of amrinone (113 +/- 33 mg three times daily), 52 patients (30%) showed a maximal increase in treadmill exercise time exceeding 2 minutes (Naughton protocol), 72 (42%) had a lesser increase, 24 (14%) developed limiting adverse reactions, 20 (12%) died and 5 dropped out of the study. Fifty-two "responders" (30%) who were free of limiting side effects and had a greater than 2 minute increase in exercise time were randomized in double-blind fashion to continued amrinone or switched to placebo (each plus standard treatment) for an additional 12 weeks. Phase 2: Comparison of 31 of these 52 responders who continued to receive amrinone with the remaining 21 randomized to placebo revealed no significant differences in vital signs, indexes of left ventricular size and function, systolic time intervals or maximal exercise time. Continued follow-up study of patients receiving either amrinone or placebo revealed decreases in exercise times of 7 and 10%, respectively (both p less than 0.05 compared with before randomization). Episodes of worsened congestive heart failure severe enough to mandate termination of double-blind treatment were as frequent in patients taking placebo (4[18%] of 21) as in those taking amrinone (4[13%] of 31; p = NS). The average symptom score and functional class of each treatment group remained comparable. Adverse effects such as gastrointestinal and central nervous system complaints were more common with amrinone treatment as were elevations of serum liver enzymes and reduced platelet counts. This large multicenter, randomized double-blind withdrawal study revealed no change in estimates of cardiac performance after the discontinuation of amrinone. These findings suggest that amrinone, in the dosages tested, does not importantly improve cardiac function beyond that provided by standard treatment with digoxin, diuretic drugs and vasodilators.