The value of a QRS scoring system derived from 12 lead electrocardiograms to estimate left ventricular ejection fraction was assessed in a prospective study of 285 hospital survivors of myocardial infarction. In these patients both the QRS score and ejection fraction were measured by radionuclide ventriculography at discharge. The correlation between ejection fraction and QRS score was weak. In 22 patients who died during six to 12 months follow up the ability of the ejection fraction and QRS score to predict mortality was assessed in terms of sensitivity, specificity, predictive value of a positive and negative test, and efficiency. For ejection fraction less than 40% and a QRS score greater than or equal to 6 sensitivity was respectively 73% and 64%, specificity 73% and 56%, predictive value of a positive test 18% and 11%, predictive value of a negative test 97% and 95%, and efficiency 73% and 56%. Both ejection fraction and QRS score may be used to identify patients at low and high risk during one year follow up, but, contrary to initial expectations, the QRS score appears to be of little value in estimating ejection fraction and is less accurate than ejection fraction in predicting late survival in hospital survivors of myocardial infarction.