We hypothesized that long-term facilitation (LTF) is due to decreased upper airway resistance (Rua). We studied 11 normal subjects during stable non-rapid eye movement sleep. We induced brief isocapnic hypoxia (inspired O(2) fraction = 8%) (3 min) followed by 5 min of room air. This sequence was repeated 10 times. Measurements were obtained during control, hypoxia, and at 20 min of recovery (R(20)) for ventilation, timing, and Rua. In addition, nine subjects were studied in a sham study with no hypoxic exposure. During the episodic hypoxia study, inspiratory minute ventilation (VI) increased from 7.1 +/- 1.8 l/min during the control period to 8.3 +/- 1.8 l/min at R(20) (117% of control; P < 0.05). Conversely, there was no change in diaphragmatic electromyogram (EMG(dia)) between control (16.1 +/- 6.9 arbitrary units) and R(20) (15.3 +/- 4.9 arbitrary units) (95% of control; P > 0.05). In contrast, increased VI was associated with decreased Rua from 10.7 +/- 7.5 cmH(2)O. l(-1). s during control to 8.2 +/- 4.4 cmH(2)O. l(-1). s at R(20) (77% of control; P < 0.05). No change was noted in VI, Rua, or EMG(dia) during the recovery period relative to control during the sham study. We conclude the following: 1) increased VI in the recovery period is indicative of LTF, 2) the lack of increased EMG(dia) suggests lack of LTF to the diaphragm, 3) reduced Rua suggests LTF of upper airway dilators, and 4) increased VI in the recovery period is due to "unloading" of the upper airway by LTF of upper airway dilators.