SUMMARY1. Central respiratory response to acute (10 min) hypoxia, as measured by phrenic nerve activity, was determined in peripheral chemo-denervated cats.2. Hypoxia was induced by ventilating cats for 10 min at reduced inspired oxygen levels (inspired 02 fraction, FI°2 = 0-06-0-15). The degree of hypoxaemia was determined from an arterial blood sample and ranged from 'severe' (arterial 02 pressure, Pa,°< 26 Torr) to 'mild' (Pa, 02 > 35 Torr). The respiratory response was monitored for 1 h following return to ventilation with 100 % oxygen.3. The results confirmed the finding of prolonged ( > 60 min) inhibition of respiration upon return to hyperoxic conditions following severe hypoxia, as reported previously (Millhorn, Eldridge, Kiley & Waldrop, 1984). A new finding was a long-lasting ( > 60 min) facilitation of respiration following exposure to less severe (Pa, 02 > 35 Torr) hypoxia.4. Medullary extracellular fluid pH was measured in six cats. Changes in pH could not explain either the prolonged inhibition following severe hypoxia or the longlasting facilitation observed following mild hypoxia.5. Ablation studies were performed in order to determine the locations of the neuronal substrates for the inhibitory and facilitatory mechanisms. The results of this series of experiments indicate that the mesencephalon is necessary for activation of the inhibitory mechanism, while the facilitatory mechanism requires the presence of higher brain structures, notably the diencephalon.6. Following removal of the diencephalon, the inhibitory response was seen following even mild hypoxic insults, i.e. those shown to produce facilitation in animals with intact brains. In the absence of the mesencephalon, neither prolonged inhibition nor prolonged facilitation could be produced following hypoxia.7. It is proposed that there are two centrally mediated long-lasting responses to acute hypoxia. Facilitation is seen following mild hypoxia. Inhibition is more likely following severe hypoxia. However, both mechanisms appear to be triggered simultaneously and the output of the central respiratory controller reflects the influence of each.