A new computer-assisted method for the quantitative assessment of the dispersion of ventricular repolarization (DVR) has been developed. Through interactive editing of an averaged QRS-T cycle from a 15-lead electrocardiographic (ECG) record (12-lead ECG + XYZ leads), five ECG indices of DVR are automatically computed: they represent the maximal interlead difference of QT and the intervals from the J point to the T wave end, from the J point to the T wave apex, and from the T wave apex to the T wave end. The standard limits of these indices were then established in six clinical groups, including normal subjects and patients with left ventricular hypertrophy, with myocardial infarction, and with intraventricular conduction defect, all subjects being without ventricular arrhythmias and without interacting drugs. The mean values and percentile ranges of all DVR indices were lower in the normal group than in all pathologic groups. The 97.5th percentiles of the QT end dispersion and the JT end dispersion were, respectively, 65 and 76 ms in normal subjects, 84 and 86 ms in patients with inferior MI; 89 and 100 ms in those with anterior MI; 90 and 98 ms in those with left ventricular hypertrophy; and 94 and 99 ms in those with intraventricular conduction defects. This suggests that increased DVR is associated with the varieties of heart disease represented in this study, even in the absence of ventricular arrhythmias, and also that individual measurements of DVR used as predictors of future arrhythmic events should be referred to the standard range of their own clinical group.