BACKGROUND: The determinants of prognosis in patients with heart failure and preserved ejection fraction (HF-PEF) are poorly documented. METHODS AND RESULTS: We evaluated data from 4128 patients in the I-PRESERVE trial (Irbesartan in Heart Failure with Preserved Ejection Fraction Study). Multivariable Cox regression models were developed using 58 baseline demographic, clinical, and biological variables to model the primary outcome of all-cause mortality or cardiovascular hospitalization (1505 events), all-cause mortality (881 events), and HF death or hospitalization (716 events). Log N-terminal pro-B-type natriuretic peptide, age, diabetes mellitus, and previous hospitalization for HF were the most powerful factors associated with the primary outcome and with the HF composite. For all-cause mortality, log N-terminal pro-B-type natriuretic peptide, age, diabetes mellitus, and left ventricular EF were the strongest independent factors. Other independent factors associated with poor outcome included quality of life, a history of chronic obstructive lung disease, log neutrophil count, heart rate, and estimated glomerular filtration rate. The models accurately stratified the actual 3-year rate of outcomes from 8.1% to 59.9% (primary outcome) 2.7% to 36.5% (all-cause mortality), and 2.1% to 38.9% (HF composite) for the lowest to highest septiles of predicted risks. CONCLUSIONS: In a large sample of elderly patients with HF and preserved EF enrolled in I-Preserve, simple clinical, demographic, and biological variables were associated with outcome and identified subgroups at very high and very low risk of events.