Two observers used two-beam pulsed Doppler ultrasound, equipped with a suprasternal probe, to measure cardiac output (QtDopp) in 38 ICU patients who had pulmonary artery catheters and in 20 adult volunteers. The two-beam pulsed Doppler method enables one device to measure simultaneously both aortic blood velocity and aortic diameter. Each observer was blind to the other's measurement and to the thermodilution cardiac output measurement (Qttd). Linear regression of the mean of both observer's QtDopp on Qttd showed QtDopp = 0.90.Qttd + 0.01 (see = 1.54 L/min, r = .90). Bias (+/- SD), defined as mean (QtDopp - Qttd) difference, was -0.69 +/- 1.55 L/min. Interobserver agreement was more variable in patients than volunteers; mean (observer 1 - observer 2) difference was 0.14 +/- 1.30 L/min in ICU patients and -0.09 +/- 0.92 L/min in volunteers. Two-beam pulsed Doppler ultrasound is a simpler method of measuring QtDopp than previous pulsed Doppler methods which measure separately the aortic diameter by echocardiography. Although its agreement with Qttd is close to other Doppler methods and has acceptable interobserver reproducibility, its accuracy remains operator-dependent.