Using a forced oscillation technique, we measured the resistance (Rrs) and reactance (Xrs) of the respiratory system between 2 and 32 Hz at three different lung volumes in 15 healthy subjects and 7 patients with chronic obstructive pulmonary disease. Rrs and Xrs were partitioned, by means of a pressure recording in the esophagus, into the resistance and reactance of lung and airways (L) and the chest wall. The measurements were validated by checking the adequacy of the frequency response of the esophagus, by the lack of difference between thoracic and mouth flow, by an estimation of the error introduced by the shunt impedance of the cheeks, and by comparisons with the values of pulmonary compliance and resistance determined in the same subjects with classical techniques. In both healthy subjects and patients, the chest wall has a low resistance that increases somewhat at low lung volumes and behaves functionally as a two-compartment system, with low capacitance at frequencies exceeding 4 Hz. Rrs varies with lung volume and is markedly frequency dependent in patients; both phenomena are due primarily to corresponding variations of RL. In healthy subjects, at and above functional residual capacity (FRC) level, the lungs behave as a one-compartment system, the reactance of which is mainly determined by the gaseous inertance, at least beyond 2 Hz. In patients and in healthy subjects breathing below FRC, the observed frequency dependence of resistance and the simultaneous increase in resonant frequency can be simulated satisfactorily by Mead's two-compartment model, assuming a large increase in peripheral airways resistance.