PURPOSE OF INVESTIGATION - Bioimpedance cardiography has been suggested as a non-invasive means to monitor cardiac function but has not been tested in cases of severe ventricular dysfunction. This study compared thermodilution stroke volume, ventriculographic left ventricular ejection fraction, bioimpedance stroke volume, and the maximum first derivative of the bioimpedance signal dZ/dtmax, during the development of experimental congestive heart failure. DESIGN - Simultaneous thermodilution stroke volume, ventriculography, and bioimpedance measurements were serially measured in pigs following acute pacing, and after 1, 2, and 3 weeks of tachycardia. Thermodilution stroke volume measurements were obtained by positioning a thermistor tipped catheter into the pulmonary artery and integrating the thermodilution curve with respect to heart rate. Left ventricular stroke volume and ejection fractions were measured from single plane ventriculograms using the area-length method. Using a series of electrodes positioned on the thoracic segment and a low level current (2.5 ma), the bioimpedance waveform was recorded and stroke volume and dZ/dtmax computed. SUBJECTS - The subjects were eight pigs (23-30 kg) with developing ventricular dysfunction due to chronic rapid atrial pacing (240 beats.min-1) and four controls. MEASUREMENTS AND MAIN RESULTS - Left ventricular ejection fraction decreased significantly from acutely paced values following 7 d tachycardia [60(SEM 1)% v 41(3)% respectively, p less than 0.01] and continued to decline with longer durations of tachycardia. A significant correlation was observed between ejection fraction and dZ/dtmax (r = 0.74, n = 32). Thermodilution and bioimpedance stroke volumes fell significantly from acutely paced values after week 2 of tachycardia [thermodilution: 13.8(0.9) v 8.5(1.4) ml; bioimpedance: 13.6(1.1) v 11.2(1.5) ml respectively, p less than 0.05] and were highly correlated throughout the study period (r = 0.90, n = 32). However, bioimpedance overestimated thermodilution values at week 2 (p less than 0.05) and at week 3 of tachycardia [thermodilution: 8.4(0.8) ml v bioimpedance: 9.6(1.0) ml, NS]. CONCLUSION - In a tachycardia induced model of heart failure, bioimpedance was significantly correlated with thermodilution stroke volume. The peak first derivative of the bioimpedance signal dZ/dtmax may provide a non-invasive index of ventricular pump performance. While these results are promising, further studies are required to evaluate the diagnostic value of bioimpedance cardiography in the clinical setting.