Left ventricular ejection fraction was assessed by biplane cross sectional echocardiography in 65 patients with a first acute myocardial infarction on the first day. In 30 patients (group 1) measurements were repeated on the third day and in another 35 patients (group 2) at three months. Changes in ejection fraction of 0.05 or less were arbitrarily called insignificant. In group 1 only two patients showed a decrease of more than 0.1 between days 1 and 3, and both had an enzymatically confirmed infarct extension. The remaining patients had no complications. In group two 11 patients had decreases of more than 0.1 between day 1 and three months: three of them had an enzymatically confirmed reinfarction (perioperative in one) and four a possible reinfarction, and in two an angiographically confirmed left ventricular aneurysm developed. In two no complications occurred. The other complications that occurred were an enzymatically confirmed but small reinfarction, an angiographically confirmed but circumscript aneurysm, and an uncomplicated bypass operation in one patient each. These three patients had a small increase (between 0.05 and 0.1) in ejection fraction. Reproducibility of the method of measuring the ejection fraction was assessed concurrently in 20 outpatients with a previous myocardial infarction who were studied twice on the same day (with a 30 minute interval) by two different observers. The mean absolute difference in ejection fraction between the paired observations was 0.036 +/- 0.023 with a range of 0 to 0.07. Thus only changes in ejection fraction of more than 0.1 correlate with clinically recognised complications. Changes between 0.05 and 0.1 may be due to spontaneous variability or to the limited reproducibility of the method.