We determined the causes of central apnea that commonly follow the hyp
erpnea resulting from brief airway occlusion during non-rapid-eye-move
ment (NREM) sleep. Ventilation and end-tidal gases were measured befor
e, during, and after 214 trials of 15-20 s of tracheal occlusion in th
ree dogs during NREM sleep. Airway occlusion was accompanied by progre
ssive increases in inspiratory effort and was followed by transient on
e- to four-breath hyperpneas, with subsequent central apnea [3-15 time
s eupneic control expiratory duration (TE)] in 62% of occlusion trials
. Significant TE prolongation after hyperventilation did not occur unt
il tidal volume (VT) was three times greater than control; i.e., there
was a volume-dependent apneic threshold. Transient electroencephalogr
am arousal at the end of the occlusion often augmented VT, thereby con
tributing to the subsequent central apnea; however, arousal was not re
quired for the apnea to occur. Significant transient hypocapnia (up to
-12 Torr arterial PCO2) commonly occurred after release of airway occ
lusion but was not closely correlated with the length of central apnea
. During vagal blockade, after release of airway occlusion, significan
t transient hyperventilation occurred but at VT < 40% greater than con
trol, and TE prolongation was markedly reduced. In summary, after rele
ase of airway occlusion in NREM sleep, 1) VT greater than three times
eupnea was necessary to cause central apnea, 2) transient arousal at t
he termination of airway occlusion caused longer apneas by augmenting
VT, and 3) transient hypocapnia per se made a significant but minor co
ntribution to the postocclusion central apnea.