We have compared the prognostic value of a predischarge post-infarction QRS score derived at rest and at submaximal exercise with ejection fraction measured by gated radionuclide left ventriculography in 65 patients. Seventeen patients died or had heart failure (group 1) and 48 were well or had angina (group 2) six months after infarction. The mean QRS score derived from the resting electrocardiogram for group 1 was significantly greater than that for group 2 (P less than 0.01) but the QRS scores at peak exercise did not differ significantly between the two groups. The mean ejection fraction for group 2 was significantly greater than that for group 1 (P less than 0.001). Both the rest and peak exercise QRS scores correlated weakly but significantly with ejection fraction (P less than 0.001). The QRS score at rest had a greater sensitivity and specificity in predicting cardiac death and heart failure than the QRS score at peak exercise. A sensitivity of 88% was achieved with a resting ejection fraction less than 50% and a resting QRS score greater than 4. At these values the specificities were 58% and 63% respectively. Combining the blood pressure response to exercise with the QRS score and ejection fraction improved the sensitivity of both with no loss of specificity. Therefore, the resting QRS score is comparable to ejection fraction as a predictor of serious cardiac events after infarction and the sensitivity of both may be improved by including an assessment of the blood pressure response to exercise. Because a 12-lead electrocardiogram is cheap and widely available, this QRS score may be used in risk stratification after infarction.