Patients with suspected myocardial infarction who present with ST depression have a high mortality which is not reduced by thrombolytic therapy. Despite this, there are few data on these patients. We studied the electrocardiographic and clinical characteristics of these patients, the diagnostic and prognostic value of the presenting electrocardiogram (ECG), and the reasons for the high mortality and apparent lack of thrombolytic efficacy. We studied all patients with suspected infarction admitted during 1990 with ST depression. Of the 136 patients (84 men, mean [SD] age 68  years), 74 (54%) had confirmed infarction and 73 (54%) had previous infarction. 1-year mortality was 26% for all patients, 31% for those with confirmed infarcts, and 19% for those in whom infarction was subsequently excluded. Patients with infarction had more severe ST depression (mean 2.5 mm [SD 1.5]) and more ECG leads with ST depression (mean 4.7 leads [1.8]) compared with patients without infarction (1.4 mm [0.8], p < 0.001; 3.6 leads [1.7], p < 0.001). Sensitivity and specificity for the subsequent diagnosis of infarction with ST depression were 20% and 97%, respectively, for at least 4 mm; and 21% and 95%, respectively, for at least 7 leads. 1-year mortality was low in patients with 1 mm ST depression (14%) or no more than 2 leads (11%), but high in patients with at least 2 mm ST depression (39%, p < 0.001) and at least 3 leads (30%, p = 0.08). Patients with suspected infarction and ST depression had a high mean age, high incidence of previous infarction, and poor prognosis. The presenting ECG is helpful in predicting prognosis, and ST depression of at least 4 mm or involving at least 7 leads is highly specific for diagnosis of infarction.