To determine whether lead adjustment formulas for correcting QT dispersion measurements are appropriate in patients after myocardial infarction. Retrospective analysis of QTc dispersion measurements in 461 electrocardiograms (ECGs). Data are presented as uncorrected QTc dispersion "adjusted" for a number of measurable leads and coefficient of variation of QTc intervals for ECGs in which between six and 12 leads had a QT interval that could be measured accurately. Patients were drawn from the placebo arm of the second Leicester Intravenous Magnesium Intervention Trial. Some 163 patients who subsequently died and an equal number of known survivors had ECGs recorded on day 2 or 3 of acute myocardial infarction. ECGs were also available in 135 of these patients from at least 1 month postinfarct. The most common lead in which a QT interval measurement was omitted was aVR (n = 176), the least common lead was V3 (n = 13). The longest QTc interval measured was most usually in lead V4 (n = 72) and the shortest in lead V1 (n = 67). As the number of measurable leads decreased there was a small, nonsignificant increase in QTc dispersion from 12 lead to eight lead ECGs (mean (SD) 100 (35.5) v 109.5 (47.9) ms). Lead adjusted QTc dispersion (QTc dispersion/square root of the number of measurable leads) showed a large, significant increase when the number of measurable leads decreased from 12 to eight (28.9 (10.3) v 38.7 (16.1) ms, P < 0.001). A similar trend was seen for coefficient of variation of QTc intervals (standard deviation of QTc intervals/mean QTc interval 64.3 (2.19) v 8.45 (3.94)%, P < 0.001). Lead adjustment formulas for QT dispersion are not appropriate in patients with myocardial infarction. Large differences in lead adjusted QTc dispersion are produced, dependent on the number of measurable leads, for very small differences in QTc dispersion. It is recommended that QT dispersion is presented as unadjusted QT and QTc dispersion, stating the mean (SD) of the number of leads in which a QT interval was measured.