A practical implementation of acoustic reflectometry for determining airway areas in routine clinical use is described. Advances over previous systems include portability, free breathing during measurements, no need to equilibrate with helium/oxygen, and real-time display of airway areas. Validation of the reflectometer with an airway model gave accuracies and reproducibilities (coefficient of variation (CV)) in the range 5-10%. With human volunteers, the within-run CV was typically 10%, and the day-to-day CV was 20%. The effect of breathing pattern on airway areas is demonstrated. In ten normal volunteers, acoustic and magnetic resonance imaging (MRI) methods of assessing pharyngeal and glottal areas were compared. The results (mean +/- SD) for the oropharynx were 1.0 +/- 0.3 cm2 acoustically and 0.9 +/- 0.5 cm2 by MRI (p = 0.77). The corresponding figures for glottal areas were 1.3 +/- 0.3 cm2 and 1.1 +/- 0.4 cm2 (p = 0.09).