We measured the velocity of audible sound at frequencies above 10 kHz to be 349 m.s-1 (+/- 1%, 95% confidence interval) in the upper airways of eupneic anesthetized dogs. It appears that the velocity of sound propagation is dependent on lumen gas composition, temperature, and mean flow speed. These three variables are sufficient to explain the velocity variations measured during tidal breathing. Lung volume, transpulmonary pressure, and bronchoactive drugs were not shown to have an appreciable effect on sound velocity. Model studies indicate that sound velocity is little affected by flow profile, lumen cross-sectional shape, or changes in lumen area with position. Flow speed proportionally changes the velocity of sound relative to the airway and therefore can be determined from sound propagation measurements. Dividing volume flow by flow speed gives airway cross-sectional area. Tracheal areas after normalization by the cube root of body weight were about half those previously reported. Acoustically inferred areas were confirmed by mechanical measurement.