Objectives:In some patients, obstructive sleep apnea (OSA) can be resolved with improvement in pharyngeal patency by sleeping lateral rather than supine, possibly as gravitational effects on the tongue are relieved. Here we tested the hypothesis that the improvement in pharyngeal patency depends on the anatomical structure causing collapse, with patients with tongue-related obstruction and epiglottic collapse exhibiting preferential improvements. Methods: Twenty-four OSA patients underwent upper airway endoscopy during natural sleep to determine the pharyngeal structure associated with obstruction, with simultaneous recordings of airflow and pharyngeal pressure. Patients were grouped into three categories based on supine endoscopy: Tongue-related obstruction (posteriorly located tongue, N = 10), non-tongue related obstruction (collapse due to the palate or lateral walls, N = 8), and epiglottic collapse (N = 6). Improvement in pharyngeal obstruction was quantified using the change in peak inspiratory airflow and minute ventilation lateral versus supine. Results: Contrary to our hypothesis, patients with tongue-related obstruction showed no improvement in airflow, and the tongue remained posteriorly located while lateral. Patients without tongue involvement showed modest improvement in airflow (peak flow increased 0.07 L/s and ventilation increased 1.5 L/min). Epiglottic collapse was virtually abolished with lateral positioning and ventilation increased by 45% compared to supine position. Conclusions: Improvement in pharyngeal patency with sleeping position is structure specific, with profound improvements seen in patients with epiglottic collapse, modest effects in those without tongue involvement and-unexpectedly-no effect in those with tongue-related obstruction. Our data refute the notion that the tongue falls back into the airway during sleep via gravitational influences.