Depressed heart rate variability (HRV) has been shown to be a powerful and independent risk factor in patients following acute myocardial infarction (AMI). A detailed comparison of the predictive values between short- and long-term HRV has not been made. The predictive value of short-term HRV for 1-year total cardiac mortality was studied in 700 consecutive patients after AMI. All patients underwent 24-hour Holter monitoring before discharge from the hospital (5 to 8 days after AMI) and were followed up for 1 year. Short-term HRV was computed as the standard deviation of all normal RR intervals (SDNN) from a 5-minute stationary period selected from 24-hour Holter electrocardiographic recordings. Long-term HRV was computed as an HRV index over the entire 24 hours. There was a significant but relatively poor correlation between SDNN and HRV index (r = 0.51, p <0.001). The positive predictive accuracy of SDNN for 1-year mortality (13% to 18%) was lower than the HRV index (17% to 43%) over a range of sensitivity of 25% to 75%. Assessment of HRV index in > or = 35% of the patients preselected by the lowest SDNN was able to achieve predictive power similar to that of HRV index assessed in all the patients. These data suggest that lower predischarge short-term HRV is associated with increased 1-year total cardiac mortality in patients after AMI. Analysis of long-term HRV for postinfarction risk stratification can safely be limited to patients preselected by depressed short-term HRV measures.