To demonstrate the role of extrathoracic airways in the overestimation of lung volumes by body plethysmography in airway obstruction, we induced in six normal subjects an artificial airway stenosis by inflating a balloon in the lower trachea. We measured thoracic gas volume in a body plethysmograph simultaneously from mouth pressure (Pm) vs. plethysmographic volume (Vbox) (TGVm) and from esophageal pressure (Pes) vs. Vbox (TGVes). During control measurements there was no difference between TGVm and TGVes (P greater than 0.1) in any subject. Balloon inflation (B) resulted in a significant (P less than 0.01) decrease of specific airway conductance (sGaw) from a mean value of 0.199 +/- 0.66 (SD) to 0.025 +/- 0.01 cmH2O-1 . s-1. TGVm significantly (P less than 0.02) exceeded TGVes in every subject (mean diff 1.24 +/- 0.56 liters) and Pm lagged behind Pes. In two subjects studied at different levels of airway obstruction we found a significant (P less than 0.05) correlation between TGVm-TGVes difference and sGaw. Bypassing the extrathoracic airways by a cuffed endotracheal tube resulted in each case in a degree of airway obstruction comparable to that during balloon inflation (sGaw 0.019 +/- 0.004 cmH2O-1 . s-1), but no difference between TGVm and TGVes (P greater than 0.1). We conclude that lung volume is overestimated by plethysmography in airway obstruction. The overestimation is due to the presence of a compliant extrathoracic airway, and it probably depends on both the degree of airway obstruction and the value of compliance of the extrathoracic airway.