Current methods for testing upper airway (UA) collapsibility in humans tend to produce intervention-related changes in some of the variables that affect UA stability. Therefore, their results may not reflect UA stability under the experimental conditions of interest. In the proposed method, the subject lies in a body enclosure with head and neck out. Pressure is altered in brief (approximately 0.2-s) pulses to avoid behavioral responses. The collapsibility of UA under "static" conditions is tested by delivering identical pressure pulses simultaneously to the airway and body surface inside the shell. Because the pressure applied to the respiratory system is not altered, cessation of flow indicates closure, and the pressure at which this happens is Pclosure. Collapsibility under dynamic conditions is tested by applying brief negative pulses to the shell only, thereby forcing an increase in inspiratory flow. Ten normal awake subjects were tested. None of the subjects developed closure when negative pulses (0 to -16 cmH2O) were applied to both airway and shell during inspiration or expiration with either nose or mouth breathing. There were only small reductions in flow, indicating minor narrowing. By contrast, pressure pulses of similar magnitude applied to the shell alone were associated with closure in 5 of 10 subjects. We conclude that the UA of normal awake humans is fairly stable under the influence of statistically applied pressure but susceptible to collapse under conditions of increased flow. Pclosure determined under static conditions underestimates the vulnerability of the UA to collapse under dynamic conditions.