The purpose of this study was to describe the technique proposed to measure left ventricular ejection fraction (LVEF) with the impedance cardiogram and to compare these values with those measured by radionuclide angiocardiography. Characteristics (mean +/- SE) of the healthy control group were: age, 32 +/- 3 yr; weight, 75 +/- 6 kg; and height, 177 +/- 3 cm. Characteristics of the patient population of 46 men and 49 women were: age, 63 +/- 1 yr; weight 74 +/- 2 kg; and height, 170 +/- 1 cm. LVEF was measured by impedance (ZEF) and multiple-gated scans (MEF) while in the supine position. The control group ZEF averaged 72% (range 67% to 78%) and the MEF averaged 71% (range 65% to 77%). There were no differences between the average ZEF (56 +/- 1%) and MEF (53 +/- 2%) in the patients. Correlations, however, between ZEF and MEF were unacceptably low for the several clinical populations within this group (-0.17 to 0.16). Furthermore, MEF correlated well with regional wall motion (r = .84) while ZEF did not (r = .00). Subdividing the patients according to heart function as determined by regional wall motion failed to improve the correlation between MEF and ZEF. The use of a previously published regression equation to predict LVEF from the systolic time interval ratio of pre-ejection period/left ventricular ejection time derived from the impedance cardiogram also proved ineffective. These data suggest that the previously proposed analysis of the impedance cardiogram to measure LVEF should not be used to make a clinical diagnosis.