The purpose of this was to determine whether the site of physiological narrowing within the upper airway was uniform or differed among patients with obstructive sleep apnea. Inspiratory pressures were measured with an esophageal balloon catheter and three catheters located at different sites along the upper airway: supralaryngeal airway, oropharynx, and nasopharynx. Peak inspiratory pressure differences between catheters allowed assessment of pressure gradients across three airway segments: lungs-larynx-retroepiglottal airway (esophageal-supralaryngeal pressure), hypopharynx (supralaryngeal-oropharynx pressure), and transpalatal airway (oropharynx-nasopharynx pressure). In five patients, hypopharyngeal obstruction was present, and in four patients no hypopharyngeal obstruction existed. In these four patients the site of obstruction was located at the level of the palate. In a given subject, the site of obstruction was the same during repeated measurements. The presence or absence of hypopharyngeal narrowing during sleep was not predictable from gradients measured across different segments of the upper airway during wakefulness. We conclude that the site of physiological upper airway obstruction varies among patients with obstructive sleep apnea and is not predictable from pressure measured during wakefulness. We speculate that uvulopalatopharyngoplasty may not relieve obstructive apneas in patients with hypopharyngeal obstruction.