Calculations of the spontaneous variability of ventricular arrhythmias are usually based upon the results of Holter electrocardiograms recorded either successively or separated by a short time interval. Only recently was it shown that the variability of ventricular premature contractions increases with longer intervals. This study was undertaken to investigate the variability of simple and complex ventricular arrhythmias over long periods to derive efficacy criteria for long-term antiarrhythmic therapy. In a prospective study, the influence of the length of the time interval on spontaneous variability was investigated in 100 patients with coronary artery disease or idiopathic dilated cardiomyopathy and untreated ventricular arrhythmia Lown grade IV. Patient follow-up was carried out for 260 +/- 387 days. In each of the 498 ambulatory Holter tapes, the mean hourly arrhythmia count (AC) of ventricular premature contractions, couplets, and salvos was verified. The variability of arrhythmia counts between two Holter electrocardiograms was defined as the logarithm of the ratio of (ACday 2 + 0.01) to (ACday 1 + 0.01). The 95% intervals for these ratios were calculated as +/- 2 SD, considering the fact that all mean values did not differ significantly from zero. The lower limit of these intervals refers to the reduction that is required for assuming drug efficacy, whereas the upper limit refers to an aggravation. The 95% intervals were calculated for each of four ranges of control intervals (0-6, 7-89, 90-364, and greater than or equal to 365 days). They increased significantly with longer control intervals.(ABSTRACT TRUNCATED AT 250 WORDS)