The safety and efficacy of long-term oral milrinone therapy were evaluated over a 2 1/2 year period in 100 patients who had severe congestive heart failure despite conventional therapy. Long-term oral milrinone therapy (27 +/- 8 mg/day initial dose) was well tolerated; drug-related side effects occurred in only 11% of patients and led to drug withdrawal in only 4% of patients. Of 94 patients evaluated after 1 month of therapy, 51% had improved by at least one New York Heart Association functional class. Despite hemodynamic and clinical improvements, life table analysis showed a 39% mortality rate at 6 months and a 63% mortality rate at 1 year of therapy. Characteristics at study entry that predicted death within 6 months included more advanced functional class, impaired renal function, lower right ventricular ejection fraction, presence of nonsustained ventricular tachycardia on 24 hour ambulatory electrocardiography, more impaired baseline hemodynamic function and absence of clinical improvement after 1 month of milrinone therapy. Multivariate analysis selected lower baseline cardiac index and aortic systolic pressure as the most significant variables in predicting death; patients who died of progressive heart failure had less frequent use of antiarrhythmic drugs and greater increases in furosemide and milrinone doses during long-term follow-up than did those who died suddenly. Thus, although milrinone is well tolerated and produces early symptomatic benefits in approximately half of patients with congestive heart failure refractory to conventional therapy, there is no evidence that it improves the high baseline mortality in this disorder.