This study considers 2 principal questions: (1) the degree to which objective application of criteria from conventional simple electrocardiographic wave measurements can approach clinical diagnosis of myocardial infarction; and (2) the degree of independent diagnostic contribution made by serial electrocardiographic change. Systematic criteria for serial electrocardiographic change do not exist. Little is known about diagnostic information contained in serial data versus that in single records. A comparison is made of electrocardiographic values measured over time in putatively healthy male control subjects with those measured before, during and after documented myocardial infarction in men in long-term studies of general populations. There was little opportunity for bias in selection of the cases of infarction from these populations, and diagnosis was based on hospital documentation during the acute phase. A systematic computer search was made for electrocardiographic criteria giving maximal sensitivity for cases of clinical infarction while holding the false positive rate in normal subjects to less than 5 percent. Analysis gave the best discriminant criteria between normal subjects and patients with infarction on the basis of (1) measurements in single post-event records, (2) change between pre- and post-event records, (3) their combination. Serial electrocardiographic changes were also weighted according to the absolute wave amplitude. The most sensitive conventional criteria for diagnosis of old anterior infarction were based on absolute R wave amplitude plus loss of R amplitude in selected leads. Sensitivity was 70 to 80 percent for diagnosis of old anterior infarction using very simple criteria, 50 to 60 percent for old inferior infarction, and 30 to 50 percent for intermediate syndromes called "coronary insufficiency." Discriminant function analysis improved the diagnosis of old infarction and coronary insufficiency. However, in this comparison, among survivors of documented myocardial infarction, the objective application of criteria for serial electrocardiographic change between pre- and post-event records made only a modest independent contribution to diagnosis of old infarction beyond that obtained from measurements in the single electrocardiogram post-infarction. The contribution of serial change might better be studied over the longer run. Many biological measurements are more effective as predictors over time than as diagnostic discriminants at a given moment.