1. The respiratory response to isocapnic hypoxia (inspired O2 fraction (FI,O1), 0.1-0.12) was measured in twelve vagotomized, paralysed, artificially ventilated young rabbits (aged 26.6 +/- 0.4 days), following pre-collicular decerebration. Phrenic nerve efferent activity was used as an index of central respiratory output (RO). In hypoxia RO increased after 1-2 min (phase 1) but decreased over the subsequent 3-4 min to, or below, the pre-hypoxic control level (phase 2). 2. We used electrical stimulation to target areas in the mesencephalon which inhibit RO. Profiles of the response to stimulation were determined in a grid of electrode penetrations made mediolaterally and rostrocaudally at the level of the superior colliculi, in normoxia. Histology confirmed that stimulation in the red nucleus (RN) inhibited RO profoundly. 3. Electrolytic lesions were made bilaterally in RN inhibitory sites or in adjacent areas. The respiratory response to isocapnic hypoxia was measured again post-lesioning. 4. In six rabbits with bilateral lesions in the RN, phase 2 of the respiratory response was abolished and RO remained elevated throughout the hypoxic exposure. However, in six rabbits with unilateral lesions in the RN, or with bilateral lesions placed in areas outside the RN that did not inhibit RO on electrical stimulation, the respiratory response remained biphasic. 5. In both groups of animals, blood pressure increased during 1-3 min of hypoxia before decreasing to pre-hypoxic levels. This cardiovascular response remained biphasic irrespective of whether animals showed a biphasic respiratory response or a sustained increase in RO after lesioning. 6. We conclude that structures within the RN are crucial to the mechanism producing a fall in RO during isocapnic hypoxaemia in the neonate.