Our study was concerned with the effect of brain hypoxia on cardiorespiratory control in the sleeping dog. Eleven unanesthetized dogs were studied; seven were prepared for vascular isolation and extracorporeal perfusion of the carotid body to assess the effects of systemic [and, therefore, central nervous system (CNS)] hypoxia (arterial = 52, 45, and 38 Torr) in the presence of a normocapnic, normoxic, and normohydric carotid body during non-rapid eye movement sleep. A lack of ventilatory response to systemic boluses of sodium cyanide during carotid body perfusion demonstrated isolation of the perfused carotid body and lack of other significant peripheral chemosensitivity. Four additional dogs were carotid body denervated and exposed to whole body hypoxia for comparison. In the sleeping dog with an intact and perfused carotid body exposed to specific CNS hypoxia, we found the following.1) CNS hypoxia for 5–25 min resulted in modest but significant hyperventilation and hypocapnia (minute ventilation increased 29 ± 7% at arterial = 38 Torr); carotid body-denervated dogs showed no ventilatory response to hypoxia.2) The hyperventilation was caused by increased breathing frequency. 3) The hyperventilatory response developed rapidly (<30 s). 4) Most dogs maintained hyperventilation for up to 25 min of hypoxic exposure. 5) There were no significant changes in blood pressure or heart rate. We conclude that specific CNS hypoxia, in the presence of an intact carotid body maintained normoxic and normocapnic, does not depress and usually stimulates breathing during non-rapid eye movement sleep. The rapidity of the response suggests a chemoreflex meditated by hypoxia-sensitive respiratory-related neurons in the CNS.