Exp Physiol 95.4 pp 561–568 561 Experimental Physiology – Research Paper Modelflow estimates of cardiac output compared with Doppler ultrasound during acute changes in vascular resistance in women Kenneth S. Dyson 1 , J. Kevin Shoemaker 2 , Philippe Arbeille 3 and Richard L. Hughson 1 1 Faculty of Applied Health Sciences, University of Waterloo, Waterloo, Ontario, Canada 2 School of Kinesiology, University of Western Ontario, London, Ontario, Canada 3 Unit´ e M´ edecine et Physiologie Spatiale, Centres Hospitaliers Universitaires Trousseau, 37044 Tours, France We compared Modelflow (MF) estimates of cardiac stroke volume (SV) from the finger pressure- pulse waveform (Finometer R  ) with pulsed Doppler ultrasound (DU) of the ascending aorta during acute changes in total peripheral resistance (TPR) in the supine and head-up-tilt (HUT) postures. Twenty-four women were tested during intravenous infusion of 0.005 or 0.01 μg kg −1 min −1 isoprenaline, 10 or 50 ng kg −1 min −1 noradrenaline and 0.3 mg sublingual nitroglycerine. Responses to static hand-grip exercise (SHG), graded lower body negative pressure (LBNP, from −20 to −45 mmHg) and 45 deg HUT were evaluated on separate days. Bland–Altman analysis indicated that SV MF yielded lower estimates than SV DU during infusion of 0.01 μg kg −1 min −1 isoprenaline (SV MF 92.7 ± 15.5 versus SV DU 104.3 ± 22.9 ml, P = 0.03) and SHG (SV MF 78.8 ± 12.0 versus SV DU 106.1 ± 28.5 ml, P < 0.01), while larger estimates were recorded with SV MF during −45 mmHg LBNP (SV MF 52.6 ± 10.7 versus SV DU 46.2 ± 14.5 ml, P = 0.04) and HUT (SV MF 59.3 ± 13.6 versus SV DU 45.2 ± 11.3 ml, P < 0.01). Linear regression analysis revealed a relationship (r 2 = 0.41, P < 0.01) between the change in TPR from baseline and the between-methods discrepancy in SV measurements. This relationship held up under all of the experimental protocols (regression for fixed effects, P = 0.46). These results revealed a discrepancy in MF estimates of SV, in comparison with those measured by DU, during acute changes in TPR. (Received 5 October 2009; accepted after revision 11 January 2010; first published online 15 January 2010) Corresponding author R. L. Hughson: Faculty of Applied Health Sciences, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1. Email: hughson@uwaterloo.ca The Modelflow method of computing cardiac stroke volume from an analysis of the pulse wave contour of the arterial blood pressure wave (Wesseling et al. 1993) has become commercially accessible with the Finometer R  device. Several previous investigations have concluded that during supine rest, head-up tilt or exercise, the Modelflow approach correlates well, with little or no discrepancy in the estimation of mean values (Harms et al. 1999; Houtman et al. 1999; Sugawara et al. 2003; van Lieshout et al. 2003), although a reference standard is required for quantitative measurements (Harms et al. 1999). Recently, stroke volume obtained with Modelflow (SV MF ) was compared with stroke volume by Doppler ultrasound (SV DU ) to reveal the correspondence of beat- to-beat changes by the two approaches (van Lieshout et al. 2003). This study showed no discrepancy between the two methods for the beat-to-beat variability in the supine posture, but did show an offset of about 10%, with the Modelflow estimates being greater than ultrasound, in a 30 deg head-up-tilt (HUT) position. The authors speculated that possible limitations in each method, including a change in heart position that could have affected the ultrasound measurements (van Lieshout et al. 2003), might have contributed to this discrepancy, which was greater than the difference observed between thermodilution and model flow at the same tilt angle (Harms et al. 1999). However, the estimation of SV MF has underlying assumptions used to derive the interrelationships between aortic characteristic impedance (Z o ), arterial Windkessel compliance (C w ) C  2010 The Authors. Journal compilation C  2010 The Physiological Society DOI: 10.1113/expphysiol.2009.050815