The aim of this study was to investigate the usefulness of heart rate variability (HRV) in predicting cardiac mortality in patients with a recent acute myocardial infarction at high risk of events. Time- and frequency-domain HRV was assessed on 24-hour pre-discharge Holter recording of 81 patients (mean age 65 +/- 10 years, 76.5% males) with 1) a recent acute myocardial infarction, and 2) left ventricular ejection fraction < 40% and/or a number of premature ventricular beats > or = 10/hour. There were 15 total cardiac deaths (18.5%) in a follow-up time of 29 +/- 15 months (range 6 to 48 months), with sudden death occurring in 11 patients (13.6%). Median values of low frequency (LF) were lower in patients with, compared to those without, total (p = 0.04) and sudden (p = 0.02) cardiac death. Similarly the low frequency/high frequency (LF/HF) ratio was lower in patients with fatal events (p = 0.03 and p = 0.02, respectively). Furthermore, mean of the standard deviations of all RR intervals for all 5-min segments < 20 ms, among time-domain variables, and very-low frequency (VLF) < 18 ms, among frequency-domain variables, significantly predicted cardiac death (relative risk-RR 2.94, p = 0.03; and RR 3.85, p < 0.005, respectively). Furthermore, VLF < 18 ms and LF/HF ratio < 1.05 significantly predicted the occurrence of sudden death (RR 3.52, p = 0.04; and RR 3.49, p = 0.04, respectively). Thus, our data show that HRV analysis is a useful tool for identifying patients with an actual increased risk of fatal cardiac events among patients who are basically considered at high risk according to the presence of an impaired left ventricular function and/or frequent ventricular arrhythmias.