The sensitivity of whole-body electrical impedance measurements to cha
nges in the volume of total body water in 12 children undergoing haemo
dialysis has been assessed. The impedance (I) of each child was measur
ed at 20-min intervals during dialysis using the standard four-electro
de technique to apply a constant current (800 mu A, 50 kHz) between th
e wrist and ankle on the non-fistula side of the patient. The ultrafil
tration volume (U) was also recorded. A simple electrical model sugges
ts that U = aHt(2) ((1/I-o)-(1/I)), where I-o is the whole body impeda
nce at the start of dialysis, a is a constant and Ht is patient height
. No significant changes in I were measured on 4 patients undergoing d
ialysis without ultrafiltration, whereas in 8 patients undergoing ultr
afiltration and dialysis I increased. Linear regression analysis and t
he above equation gave a mean value for a = 0.566 1 Ohm/cm(2) (coeffic
ient of variation = 3%), (mean r = 0.97), values comparable to those v
alues obtained from isotope dilution studies. Predicted fluid loss in
8 patients following a single dialysis session gave a mean overestimat
e of 4.3% (limits of agreement 27.3% and -19.7%), although in 6 of the
patients agreement was to within 6%. Changes in impedance reflect cha
nges in total body water in children undergoing haemodialysis and are
relatively insensitive to factors such as the possible differences in
electrolyte levels between these patients.