The influence of lead selection on QT estimation in the 12-lead electrocardiogram was assessed in 63 patients (21 control subjects, 21 with anterior myocardial infarction, 21 with inferior myocardial infarction). QT estimates varied between leads. The variation was greater in patients with myocardial infarction than in control subjects (mean dispersion of QT: control subjects, 48 +/- 18 ms [+/- standard deviation]; anterior myocardial infarction, 70 +/- 30 ms; inferior myocardial infarction, 73 +/-32 ms). The maximum QT in any lead (QTmax) was determined and the deviation of each lead from this maximum value calculated. In all 3 groups, anteroseptal leads (V2 or V3) provided the closest approximation to QTmax. Interlead variability was found to be mainly due to variation in timing of the end of the T wave, rather than the onset of the QRS complex. The variability due to leads was considerably greater than the variability due to cycles, observers or measurement error. Implementation of a variety of current lead selection practices resulted in widely divergent estimates of QT interval.