The applicability of the finger pressure-derived pulse contour (PC) technique was evaluated in the measurement of stroke volume (SV), cardiac output (CO) and their changes in different phases of the tilt-table test. The reference method was whole-body impedance cardiography (ICG). A total number of 40 physically active patients, aged 41 +/- 19 years, were randomly chosen from a pool of 230. Specifically speaking, 20 of the patients experienced (pre)syncope (tilt+ patients) during the head-up tilt (HUT), and 20 did not (tilt-). A total number of three measurement periods, 30-60 s each, were analysed: supine position, 5 min after the commencement of HUT, and 1 min before set down. SV and CO values measured by PC underestimated significantly those measured by ICG (biases +/- SD 19 +/- 14 ml and 1.55 +/- 1.14 l min-1, respectively) in agreement with earlier reports. The bias between the methods was almost the same in the different phases of the test. However, the SD of the bias was bigger for tilt+ (P < 0.05). When the bias between the methods was eliminated by scaling the first measurement to 100%, the agreement between the methods in the second and third measurements was clearly better than without scaling. Both methods showed a physiological drop in SV after the commencement of HUT. These results indicate that PC suffices in tracking the changes in CO and SV, but for absolute values it is not reliable.