To determine the limits of agreement between left ventricular ejection fraction estimated using systolic time intervals from impedance cardiography and left ventricular ejection fraction estimated by radionuclide ventriculography. Design:A prospective study for diagnostic tests using radionuclide ventriculography as the criterion standard. Setting:A large, military teaching hospital. Patients:Twenty ambulatory adults scheduled for radionuclide ventriculography. Measurements and Main Results:A regression equation to estimate ejection fraction from systolic time intervals is available in a widely used impedance-based cardiac monitoring device. The estimated ejection fraction is then used in an equation with stroke volume estimated by the same device to calculate an end-diastolic volume. We studied the agreement of the ejection fraction as estimated by this device with the ejection fraction estimated by radionuclide ventriculography by obtaining simultaneous estimates of ejection fraction over a broad range of adult patients.Twenty ejection fraction pairs were analyzed. The correlation of ejection fraction by impedance cardiography to ejection fraction by radionuclide ventriculography was significant (r2 = .55; p < .002). However, the mean difference between the technologies was −8.85%, with a standard deviation of the differences of 7.15%, resulting in a 95% confidence range for agreement of −23.2% to +5.5%. Conclusions:The 95% confidence range defining the limits of agreement between ejection fraction by impedance cardiography and ejection fraction by radionuclide ventriculography is not clinically acceptable. In the opinion of the authors impedance cardiography should not be used in place of radionuclide ventriculography for the assessment of ejection fraction at this time. (Crit Care Med 1993; 21:1523–1527)