To study the relations of central venous flow velocity (VFV) to cardiac pump function, hepatic venous flow velocity was recorded using Doppler echocardiography in six normal subjects and 47 patients with heart disease, of whom 28 had sinus rhythm and 19, atrial fibrillation. The area under the VFV profile during systole and diastole in a cardiac cycle was computed, and termed the VIs (systolic time-velocity integral) and VId (diastolic time-velocity integral), respectively. VIs was divided by the sum of VIs and VId [VIs/(VIs + VId)], and this was termed the VI ratio. The cardiac index (CI) was estimated by Doppler echocardiography. In normal subjects, the VFV pattern in a cardiac cycle was biphasic, the systolic VFV being dominant. In patients with atrial fibrillation, the systolic VFV was attenuated or absent, the diastolic VFV being dominant. The CI correlated well with the VI ratio (r = 0.80; p less than .001) in 14 patients with atrial fibrillation except for five patients with tricuspid regurgitation. Four patients in whom atrial fibrillation converted spontaneously to sinus rhythm showed an increase in the CI and the VI ratio according to the CI-VI ratio relationship. In patients with sinus rhythm, the CI tended to decrease as the VI ratio increased. In 15 patients who had a VI ratio of over 0.75, the CI correlated inversely with the VI ratio (r = -0.70; p less than 0.01). Three of four patients who had the VI ratio of 1.0 died of congestive heart failure. Although there was positive correlation between the CI and VI ratio in patients without effective atrial contraction, there was inverse correlation in patients with effective atrial contraction. It is suggested that the VI ratio could be a good indicator of cardiac pump function.