The upper airway wall impedance (Zuaw) may be responsible for a large artifact in the measurement of respiratory system impedance (Zrs) in children. In 17 normal children aged 3.5-13 years Zuaw and Zrs were estimated by varying transrespiratory pressure directly at the mouth (conventional method: Z1) and around the subject's head (head generator method: Z2) from 4 to 32 Hz. Zrs and Zuaw were calculated from Z1 = Zrs.Zuaw/(Zrs+Zuaw) and Z2 = Zrs (1 + Zp/Zuaw), where Zp is the impedance of the pneumotachograph. From the real and imaginary part of Z1, Z2, Zrs, and Zuaw, the corresponding resistance, inertance, compliance and resonant frequency were calculated assuming simple RIC models. No significant difference was found between the mean +/- SE of parameters derived from Zrs (respectively, 6.8 +/- 0.4 cmH2O.L-1.s, 0.034 +/- 0.001 cmH2O.L-1.s2, 10.4 +/- 0.8 m.cmH2O-1, 9.1 +/- 0.3 Hz) and Z2 (6.8 +/- 0.4 cmH2O.L-1.s, 0.038 +/- 0.002 cmH2O.L-1.s2, 10.7 +/- 0.7 ml.cmH2O-1, 8.7 +/- 0.4 Hz). All but the compliance, derived from Z1 were significantly different (P less than 0.01) from those derived from Zrs (5.3 +/- 0.3 cmH2O.L-1.s, 0.008 +/- 0.001 cmH2O.L-1.s2, 11.9 +/- 1.2 ml.cmH2O-1, and 20.3 +/- 1.6 Hz). Respiratory resistance and compliance correlated significantly with height (r = -0.56 and 0.86, respectively), in contrast to upper airway wall resistance (Ruaw) and compliance (Cuaw). Ruaw (8.6 +/- 0.8 cmH2O.L-1.s), Cuaw (1.2 +/- 0.2 m.cmH2O-1), and upper airway wall inertance (0.030 +/- 0.004 cmH2O.L-1.s2) were close to those obtained by direct measurements in adults. The mechanical properties of the upper airway wall are responsible for a significant error in the measurement of Zrs by the conventional method in normal children. Most of the artifact may be corrected for by applying pressure around the child's head.