The site of origin of the vesicular lung sound has long been in question. A technique (subtraction phonopneumography) is described here for determining the relative distance of a sound source from the chest wall. This technique involves the simultaneous recording of lung sounds from two different sites on the chest wall, phase inversion of one of the signals, and then mixing the signals in a summing amplifier. The degree of cancellation that results is inversely proportional to the number of sources and the degree to which each source is detected by both microphones simultaneously. A study of six normal subjects revealed little or no cancellation of inspiratory vesicular sounds with microphones separated by 10 cm. During expiration, cancellation did occur at distances well beyond 10 cm and was detectable over several homologous segments of opposite lungs. This finding is consistent with an intrapulmonic and probably intralobar source for the inspiratory component and an upper airway source for at least some of the expiratory component.