Heart rate variability (HRV) computed from 24-hour ECG recording has been associated with an increased risk of malignant arrhythmias after MI. To make HRV analysis more practical, we evaluated prospectively prognostic role of short-term HRV in comparison with other risk stratifiers. Study population consisted of 48 patients with acute MI (mean age 59.6 +/- 10.6 years, 38 males), who were off betablockers. All patients underwent 30-minute ECG recording at supine rest on day 2 and 5 after admission, between 9 and 11 a.m. One ECG channel from a commercial bedside monitor was A/D converted, and subsequently analysed using a purpose-built interactive software. Short-term HRV was computed as the standard deviation of all normal-to-normal RR intervals (SDNN) as well as the square root of the mean of the sum of the squares of differences between adjacent normal RR intervals (rMSSD). Left ventricular ejection fraction (LVEF, in %) was determined using 2D-echocardiography. During one-year follow up, 5 patients (10.4%) died of sudden cardiac death (SCD) and one of non-cardiac death. Subjects who died of SCD presented with significantly lower SDNN parameter on day 5 (28.8 +/- 4.3 vs 39 +/- 18.4, p < 0.006) and similar trend was revealed for rMSSD (12.22.8 vs 24.321, N.S.). Similarly, LVEF was significantly decreased in these patients (35.4 +/- 5.5 vs 49.7 +/- 11.3, p < 0.007). Positive predictive accuracy for prediction of SCD was 17% for rMSSD, 20% for SDNN, 29% for LVEF, and 40% for combination of depressed SDNN (< or = 33 ms) and LVEF (< or = 40). In conclusion, depressed HRV computed from short-term predischarge ECG recordings obtained under standardised conditions is associated with an increased risk of SCD. Such predictive power is substantially increased in combination with depressed LVEF, and this approach seems to be effective as a simple screening method to identify high risk subjects.