Total respiratory impedance was assessed from 4 to 30 Hz in ten normal subjects breathing air and a helium-oxygen gas mixture using two methods of applying pressure oscillations at the airway opening: 1) the conventional method where pressure is varied at the mouth: 2) the method recently developed by Peslin et al. (J Appl Physiol, 1985, 59, 1790-1795) in which pressure is varied both at the mouth and around the head to minimize transmural pressure across upper airway walls, and the corresponding artefact. When breathing air slightly lower resistances (p less than 0.05) and considerably higher inertances (p less than 0.001) were found using the head generator. Breathing helium-oxygen reduced respiratory resistance and its frequency dependence, as well as respiratory inertance very significantly (p less than 0.001), with minor differences between the changes seen with the two methods. In contrast, the changes in respiratory compliance were small, and not in the same direction, when pressure was varied at the mouth and around the head. It is concluded that the accuracy of the conventional method may be sufficient for diagnostic purposes in subjects without gross mechanical abnormalities, i.e. for early detection of mechanical abnormalities.