Background: Chronic heart failure (CHF) is a frequent disease with a dismal prognosis, but little is known about survival in the very elderly. There are no data on the prognostic value of cardiopulmonary exercise testing in this population. We aimed to assess exercise capacity, survival, and prognostic parameters in elderly patients with CHF. Methods: We evaluated 188 patients with CHF >70 years old (mean 77 ± 4 years, range 70-94 years) seen at our heart failure clinic between March 1992 and June 1998. A cardiopulmonary exercise test was performed in 102 patients (peak Vo2 15.3 ± 4.7, VE/Vco2 slope 39.6 ± 15.01). All patients were followed up for at least 12 months. The prognostic end point of the study was all-cause mortality. Results: At the end of follow-up (16 ± 10 mo, range 12-41 mo), 67 patients (35.6%) had died (1-year mortality rate 26% [95% confidence interval 20-32]). In univariate analysis New York Heart Association class (NYHA) (relative risk [RR] = 2.56, P < .0001), VE/Vco2 (RR = 1.041, P < .0001), peak Vo2 (RR = 0.87, P = .0007), and fractional shortening (RR = 0.95, P < .0001) predicted mortality. Peak Vo2 predicted mortality independently of age, NYHA class, and left ventricular ejection fraction. A subgroup of 12 patients with dynamic left ventricular outflow tract obstruction during stress had an excellent outcome, with a 100% survival at the end of follow-up (mean 16 ± 7 mo, range 12-39 mo). Conclusions: The prognosis in elderly patients with CHF is poor. Valid exercise testing results can be obtained in more than 50% of elderly patients with CHF. NYHA class and peak Vo2 are the strongest prognostic factors in this population.