The aim of this study was to: (1) compare the usefulness, in clinical practice, of different echocardiographic methods of left ventricular (LV) function determination in patients with a recent thrombolytic-treated acute myocardial infarction (AMI); (2) compare these measurements with the reference method radionuclide imaging; and (3) evaluate the reproducibility of visual estimation of the LV ejection fraction (EF) and the use of the biplane method of discs (Simpson's rule) in clinical practice. Echocardiography and radionuclide imaging were performed within 2 hours of each another, 5 to 8 days after hospital admission. Ninety-six patients (70 men and 26 women) age 64 +/- 9 years (range 45 to 75) were studied. The echocardiographic study was performed by 2 experienced physicians, independently of each another. LV wall motion score index and visual estimation of the EF correlated best with the radionuclide EF (r = 0.72 and r = 0.71), thereafter simply counting the number of affected LV segments (r = 0.67) or atrioventricular plane measurements (r = 0.64). Simpson's rule had low correlation to the radionuclide EF (r = 0.45 to 0.51) and could not be used in approximately half of the patients due to poor identification of endocardial borders. The interobserver coefficient of variation for independent visual echocardiographic estimation of the EF was 10%, for Simpson's rule 18%, and for the radionuclide EF 5%. We conclude that the EF estimated from quantitative echocardiographic volume calculations (Simpson's rule) may differ substantially from radionuclide methods of measuring the EF. However, with experienced sonographers, the LV wall motion score index or visual estimation of the EF had reasonable agreement with the radionuclide EF in most of the patients. Atrioventricular plane measurement is an acceptable alternative.