Bioimpedance spectroscopy (BIS) has been suggested for the assessment of fluid shifts between intracellular (ICV) and extracellular volume (ECV) during dialysis. The electrical tissue parameters are estimated by fitting a Cole-Cole model to the impedance data. Those parameters are used for the calculation of ICV and ECV with a fluid distribution model (FDM). We investigated whether postural changes cause artifacts in the volume data measured with a commercial BIS system. This is of importance at the beginning of dialysis, when the patient lies down for treatment. Volume estimations were performed during tilt table experiments with 11 healthy volunteers. Impedance spectra (5 to 500 kHz) were recorded for the total body as well as for body segments (leg and arm) during three phases: (1) 30 minutes resting in a supine position after standing; (2) 30 minutes 70 degrees head up tilt; and (3) a 30-minute resting period in a supine position. ECV and ICV were estimated with a commercially utilized FDM which is based on Hanai's mixture theory. A monoexponential function was fitted to the data for extracting the time constants and the extrapolated steady state values of the volume changes. The ECV and ICV data changed significantly during all three periods, that is, a steady state could not be reached within 30 minutes. During phase 1 the ECV decreased by 1.8 +/- 0.7%, in the tilt phase it increased by 3.8 +/- 1.1%, and in phase 3 it decreased again by 2.9 +/- 1%. The ICV increased by 3.6 +/- 2.4% during phase 1 and decreased by 6.8 +/- 5.1% during tilting; in phase 3 it increased by 4.6 +/- 1.7%. The time constants were 36.4 +/- 12.7 minutes (ECV) and 10.8 +/- 5.4 minutes (ICV) during phase 3. Segmental measurements revealed that the legs contribute significantly to the measured volume changes. The absolute volume changes in ICV and ECV differed significantly in all phases, and the same was found for the time constants during phases 1 and 3. From this discrepancy it is concluded that the measured volume changes are artifacts that are caused by extracellular fluid redistribution. Furthermore, it appears unlikely that the measured fluid shifts actually occur between ECV and ICV in the absence of osmotic changes in the body fluids. The validity of the method for a reliable assessment of volume changes during dialysis appears questionable, as dialysis-induced volume changes lie in the same range as the orthostatically-induced spurious volume changes.