QT dispersion is an easily obtained measurement from the standard electrocardiogram, and is the maximum minus the minimum QT interval. QT dispersion might represent dispersion of ventricular repolarisation, and be a marker of arrhythmic substrates. The main aim of the thesis was to explore a possible association of increased QT dispersion with death after myocardial infarction. In a preliminary study of 20 patients with acute myocardial infarction, QT dispersion was increased on day 1, increased to a maximum on day 3, and was falling by hospital discharge. A case-control study was undertaken in which 163 patients who subsequently died were drawn from the placebo am of a large randomised trial in myocardial infarction and were matched with surviving controls. Comparison of QT dispersion measured on day 2 or 3 in patients who subsequently died and those who survived showed no association between QT dispersion and mortality. However in the subset of patients for whom later electrocardiograms were available, a significant association between mortality and late QT dispersion was observed. However these patients may represent a selected population. The mechanism of increased QT dispersion remains unclear. QT dispersion measurements in patients undergoing angioplasty, or with dilated cardiomyopathy, indicated that neither acute ischaemia nor left ventricular size were satisfactory explanations for increased QT dispersion. It became apparent that there were methodological problems in QT dispersion measurement. Estimates of the reproducibility of QT dispersion in the post-infarct situation were made, and problems with automated measurement of QT dispersion identified. "Lead adjustment" formulae were shown to be inappropriate. A simplified measurement protocol was proposed. In summary, dynamic changes in QT dispersion following infarction have been confirmed, but early dispersion measurements fail to predict mortality. Failure of QT dispersion to return to normal is associated with increased mortality, but both the validity of this observation and its mechanism require further evaluation.