Background: Heart failure (HF) is associated with frequent hospitalization and worse prognosis. Prognosis factors and survival in very long-term follow-up have not been reported in HF. HF disease management programs(DMP) results are contradictory. DMP efficacy in very long-term follow-up is unknown. We studied the very long-term follow-up of up to 23.6 years and prognostic factors of HF in 412 patients under GDMT included in the REMADHE trial. Methods: The REMADHE trial was a prospective, single-center, randomized trial comparing DMP versus usual care(C). The first patient was randomized on October 5, 1999. The primary outcome of this extended REMADHE was all-cause mortality. Results: The all-cause mortality rate was 88.3%. HF was the first cause of death followed by death at home. Mortality was higher in the first 6-year follow-up. The predictive variables in multivariate analysis associated with mortality were age >52 years (P=0.015), Chagas etiology (P=0.010), LVEF <45% (P=0.008), use of digoxin (P=0.002), functional class IV (P=0.01), increase in urea (P=0.03), and reduction of lymphocytes (P=0.005). In very long-term follow-up, DMP did not affect mortality in patients under GDMT. HF as a cause of death was more frequent in the C group. Chagas disease, LVEF <45%, and renal function were associated with different modes of death. Conclusion: DMP was not effective in reducing very-long term mortality; however, the causes of death had changed. Our findings that age, LVEF, Chagas disease, functional class, renal function, lymphocytes, and digoxin use were associated with poor prognosis could influence future strategies to improve HF management.