QT dispersion may serve as a measure of variability in ventricular recovery time and may be a means of identifying patients at risk of arrhythmias and sudden death after acute myocardial infarction. We investigated this possibility on electrocardiograms (ECGs) recorded 2 or 3 days after infarction (early) and at least 4 weeks later (late). 163 patients who died between 1 day and 5 years after infarct were compared with an equal number of survivors matched for age and sex. 53 of the patients who died and 82 survivors also had late ECGs. There was no difference in early QT dispersion between the patients who died and the survivors (mean QTc dispersion 112.1 [SD 44.4] vs 109.9 [42.7] ms1/2). QTc dispersion fell significantly from early to late ECGs in survivors (110.9 [48.5] to 76.5 [28.8] ms1/2), but not in patients who died during follow-up (108.0 [51.0] to 98.9 [43.1] ms1/2). The difference between the groups in the mean change was significant (34.4 [55.2] vs 9.1 [60.8] ms1/2, p = 0.016). QT dispersion measured on an ECG recorded 2 or 3 days after acute myocardial infarction does not predict mortality during the next 5 years. Increased QT dispersion on ECGs recorded at least 4 weeks after infarct may be associated with subsequent mortality, but this finding must be confirmed in a prospective trial.