We compared central venous oxygen saturation and central venous pressure (CVP) as indices of the effective blood volume during 50 degrees head-up tilt (anti-Trendelenburg's position) induced hypovolaemic shock in eight healthy subjects. Head-up tilt increased thoracic electrical impedance from 31 (28-36) (median and range) to 34 (30-40) Ohm, mean arterial pressure (MAP) from 79 (70-88) to 86 (80-99) mmHg, heart rate (HR) from 67 (56-71) to 99 (78-119) beats min-1 (p < 0.01) and total peripheral resistance (TPR) from 20 (16-32) to 36 (15-52) mmHg min l-1 (p < 0.03). Cardiac output decreased from 4.3 (3.0-4.8) to 2.7 (1.8-4.8) l min-1 (p < 0.03). After 29 (9-56) min presyncopal symptoms appeared, together with a decrease in MAP to 63 (43-79) mmHg, HR to 68 (30-112) beats min-1 and TPR to 22 (13-33) mmHg min l-1 (p < 0.02). During tilting CVP decreased from 3 (1-6) to 1 (-3-5) mmHg (p < 0.05) but thereafter remained stable. In contrast, central venous oxygen saturation showed a linear decrease with time from 0.75 (0.69-0.78) at rest to 0.60 (0.49-0.67) (p < 0.01) when presyncopal symptoms appeared. Central venous catheterization is a tool for measurement of central venous oxygen saturation in addition to recording of CVP. The results suggest that a reduced central blood volume is reflected more clearly in central venous oxygen saturation than in CVP.