A method for estimating infarct size from 12-lead ECGs has been developed but not extensively validated. To assess its accuracy, ECG scores from 62 patients admitted to the coronary care unit at Barnes Hospital were compared to infarct size calculated from plasma MB creatine kinase (MB-CK) activity. A second cohort of 29 patients enrolled in the Multicenter Investigation of the Limitation of Infarct Size (MILIS) was evaluated as a test set and to provide pathologic correlates. Patients with conduction system disease, ventricular hypertrophy, or multiple infarctions were excluded, as were those in the Barnes group who had undergone thrombolytic therapy. ECGs obtained early (days 3 to 7 in the Barnes group and day 3 in the MILIS group) or late (days 8 to 14 in the Barnes group) were scored manually and by computer. QRS scores from early ECGs of patients with anterior infarctions correlated closely with MB-CK estimates of infarct size (r = 0.71 [Barnes] and 0.85 [MILIS]) and with anatomic data (r = 0.78). Enzymatic and pathologic infarct size also correlated well (r = 0.85). Correcting for body surface area by means of total CK-derived infarct size or use of QRS scores from late ECGs did not alter the correlation coefficients. Among patients with inferior infarctions QRS scores corresponded poorly with MB-CK infarct size (r = 0.28 [Barnes] and r = -0.42 [MILIS]) and pathologic infarct size (r = -0.20), despite a significant relationship between pathologic and MB-CK estimates (r = 0.62). Correction for body surface urea, use of total CK-derived infarct size, and use of QRS scores from late ECGs did not improve the results. Thus, the QRS scoring system provides reliable estimates of infarct size in patients with anterior infarctions but correlates poorly and may be misleading in patients with inferior infarctions.