Rapid eye movement sleep (REM) is linked to lower ventilation and more severe obstructive sleep apnea (OSA) than non-REM (nREM) sleep for unknown reasons. Researchers used direct physiological measurements to see if the pharyngeal compromise in REM OSA is due to neuronal ventilatory drive withdrawal or pharyngeal pathophysiological deficiencies in general (i.e., increased collapsibility, decreased muscle responsiveness). Sixty-three people with OSA participated in sleep studies that included gold-standard assessments of ventilatory “drive” (calibrated intra-esophageal diaphragm EMG), ventilation (oronasal “ventilation”), and genioglossus EMG (EMGgg). They used measures at nadir drive to determine drive withdrawal (1st decile drive within the stage). They used Collapsibility (reduced ventilation at eupneic drive) and responsiveness to evaluate pharyngeal physiology (ventilation-drive slope). Mixed model analysis was used to compare REM to nREM, and a sensitivity analysis was used to look at phasic REM. In 25 individuals, REM ( ≥10 min) was recorded. When compared to nREM, REM drive dropped to significantly lower nadir levels (1st decile: 21.8[31.2,12.4]% eupnea; P<0.0001, estimate[95% CI]), accompanied by a decrease in ventilation (25.8[31.8,19.8]% eupnea; P<0.0001). On the other hand, REM did not affect collapsibility (eupneic drive breathing), baseline EMGgg activity, or responsiveness. After correcting for nadir drive (+4.3[4.2,14.6]), REM was linked with greater OSA severity (+10.1[1.8,19.8] events/hr). In phasic REM, there was a worsening of drive withdrawal. Vendors drive withdrawal, rather than preferential decrements in muscle activity or response, appears to cause pharyngeal impairment in OSA patients. The primary goal of REM OSA treatment may be to prevent drive withdrawal.