Two methods of measuring ventilatory mechanical impedance (Z) by forced oscillations between 6 and 20 Hz were compared in 24 infants aged 2 to 49 months: 1) the application of pressure oscillations at the airway opening (Z1); and 2) the application of pressure oscillations around the head (Z2). The latter has been recently proposed to minimize the influence of compliant upper airway walls (Peslin et al., J Appl Physiol. 1985, 59:1790-1795). Ventilatory resistance and compliance (Rsb, Csb) were also obtained with the single breath method. The real part of Z1 (R1) was markedly lower than that of the corresponding Z2 (R2), at any frequency. R1 exhibited a systematic negative frequency dependence, in contrast with R2. At any frequency, the slope of the regression equation on Rsb was closer to unity for R2 than for R1. The imaginary part of Z1 (X1) was negative over the whole frequency interval, and negative values of inertance were derived from X1. X2 was negative at low and positive at high frequencies. Resonant frequency (mean +/- SD = 10.5 +/- 3.5 Hz) was always reached with Z2 and correlated negatively with body weight (r = -0.61). Inertance estimated from X2 was positive and correlated negatively with body height (r = -0.66). The compliance derived from Z1 (C1 = 3.35 +/- 2.32 10(-3) L.cm H2O-1) was not significantly different from that derived from Z2 (C2 = 2.99 +/- 2.02 10(-3) L.cm H2O-1). The marked difference observed between Z1 and Z2 is related to the importance of the upper airway shunt and may be explained by inaccuracies of both methods.(ABSTRACT TRUNCATED AT 250 WORDS)