Background: The aim of this study was to evaluate whether abnormalities in heart rate variability (HRV) could act as markers of ventricular tachycardia and prognosis in patients with advanced, chronic heart failure. Fifty patients with chronic heart failure (45 men; mean age, 59 +/- 9 years; New York Heart Association [NYHA] class II-III; left ventricular ejection fraction [LVEF], 19 +/- 9% and peak oxygen consumption, 16.6 +/- 5.4 mL/kg/min) caused by idiopathic dilated cardiomyopathy (n = 12) and ischemic heart disease (n = 38) were included in the study. Heart rate variability measures derived from 24-hour electrocardiographic (ECG) monitoring (Marquette 8500 recorder, Marquette Electronics, Milwaukee, WI) were calculated in the time domain and frequency domain., Methods and Results: Twenty-five patients (50%) revealed episodes of ventricular tachycardia on 24-hour ECG monitoring (1-143 episodes). The presence of ventricular tachycardia was associated with lower LVEF but there was no difference in NYHA class and peak oxygen consumption between patients with and without ventricular tachycardia (LVEF, 16 vs 22%, P = .01; NYHA class, 2.6 vs 2.4; peak oxygen consumption, 16.5 vs 16.8 mL/kg/min, not significant). Patients with ventricular tachycardia exhibited markedly lower HRV measures. Multiple regression analysis was used to test HRV parameters as potential predictors of ventricular tachycardia. Among them, high-frequency power was the only independent predictor of the presence of ventricular tachycardia, and this predictive correlation was independent of LVEF and mean R-R interval duration. During a follow-up period of 24 +/- 18 months, 12 patients (24%) died. No difference was found in age, etiology, NYHA class, peak oxygen consumption, or occurrence of ventricular tachycardia, but a lower LVEF (15 +/- 6 vs 21 +/- 9%, P = .046) was observed in those who died compared with those who survived. Certain estimates of HRV were in contrast, lower in those who subsequently died: standard deviation of all normal R-R intervals (61 +/- 30 vs 101 +/- 33 ms), standard deviation of 5-minute mean R-R intervals (55 +/- 27 vs 92 +/- 31 ms), mean of all 5-minute standard deviations of R-R intervals (22 +/- 12 vs 37 +/- 11 ms), and the low-frequency (3.2 +/- 1.8 vs 4.8 +/- 0.9 ln ms2) and high-frequency (3.0 +/- 1.1 vs 3.8 +/- 0.8 ln ms2) components of the HRV spectrum (all differences, P < .01). In univariate Cox analysis, all of these HRV measures were independent predictors of death. Kaplan-Meier survival analysis revealed that the standard deviations of all normal R-R intervals and of 5-minute mean R-R intervals dichotomized at median values (99 and 90.5 ms, respectively) were the best predictors of mortality., Conclusions: In patients with moderate to severe chronic heart failure, depressed indices of HRV on 24-hour ambulatory ECG monitoring could be related to higher risk of ventricular tachycardia and death, suggesting that analysis of HRV could be usefully applied to risk stratification in chronic heart failure patients.