To determine the reproducibility of frequency domain measures of heart period variability in patients with previous myocardial infarction, 2 random samples of 40 patients each (1 from the Cardiac Arrhythmia Pilot Study [CAPS] [unsustained ventricular arrhythmias], and 1 from the Electrophysiologic Studies Versus Electrocardiographic Monitoring [ESVEM] [sustained ventricular arrhythmias] trial) were studied. For each patient, two 24-hour continuous electrocardiographic recordings were analyzed, and the average normal RR interval, total power and 4 components of total power were calculated. Group means and standard deviations for each sample were virtually identical for the pairs of 24-hour recordings. Furthermore, measurements for individual patients were stable from day to day, as measured by the intraclass correlation coefficients and the standard errors of measurement. Reproducibility of heart period variability measurements is excellent in patients with previous myocardial infarction and ventricular arrhythmias, and is comparable to the high stability previously found in a small group of normal subjects. The stability of measures of heart period variability facilitates distinguishing real changes due to progression or regression of cardiac disease or to drug effects from apparent changes due to random variation.