A low heart rate variability (HRV) has been shown to be a powerful predictor of cardiac events in patients surviving an acute myocardial infarction (MI), but it is not clear yet which among the HRV parameters has the best predictive value. Time domain and frequency domain HRV was assessed on 24-hour predischarge Holter recording of 239 patients with a recent MI. Patients were followed up for 6 to 54 months (median 28), during which 26 deaths (11%) occurred, 19 of which were cardiac in origin and 12 were sudden. Most HRVs did not show any difference between patients with or without mortality end points, but the average low-frequency and low-frequency/high-frequency ratio was lower in patients with events. However, when dichotomized according to cut points that maximized the risk of sudden death, several HRVs were significantly predictive of clinical end points. Overall, the mean of the standard deviations of all RR intervals for all 5-minute segments and the standard deviation of the mean RR intervals for all 5-minute segments were the time domain variables most significantly associated with mortality end points, whereas very low frequency was the most predictive frequency domain variable. Compared with the best time domain variables, very low frequency showed a better sensitivity (0.27 to 0.42 vs 0.19 to 0.33) for end points with only a small loss in specificity (0.92 vs 0.96). On multivariate Cox proportional analysis, a left ventricular ejection fraction <40% and a number of ventricular premature beats > or = 10/hour were the most powerful independent predictors for all end points, whereas no HRV was independently associated with the events. A low frequency/high frequency ratio < 1.05 only had a borderline association with sudden death (RR = 2.86, p = 0.076). Our data show a strong association between HRV and mortality in patients surviving a recent MI, with a slight better sensitivity of frequency domain analysis. In our study, however, HRV did not add independent prognostic information to more classic prognostic variables (e.g., left ventricular function and ventricular arrhythmias).