1. The breathing pattern following acute hypoxia (arterial O2 pressure (Pa,O2), 27.4 +/- 7.7 mmHg) was measured in intact, anaesthetized and spontaneously breathing adult rats (n = 4) and in anaesthetized, vagotomized, paralysed and ventilated animals (n = 14). Measurements were made both before and after bilateral lesions or chemical inactivation of neurones in the lateral pons. Respiratory motor activity was recorded as an index of the respiratory cycle. We tested the hypothesis that the ventrolateral pons is required for expression of post-hypoxic frequency decline, defined as a decrease in respiratory frequency below steady-state baseline levels following brief exposures to hypoxia. 2. We identified an area in the ventrolateral pons where brief (1 ms) low current (< or = 20 microA) pulses evoked a short-latency inhibitor of phrenic nerve activity. At this site, bilateral electrical or chemical lesions (n = 3) were performed, or neural activity was inhibited by focal injections of 10 mM muscimol (n = 9). In six control animals, neural activity was inhibited by muscimol injections into the lateral pons, dorsal to the target site. 3. Prior to pontine intervention, respiratory frequency decreased below baseline levels following 20-110 s of 8% O2. The decrease in frequency resulted from a prolongation of expiration (up to 276%), which gradually returned to baseline levels (tau = 45 s). 4. Following lesions or inhibition of neural activity in the ventrolateral pons, baseline inspiratory (TI) and expiratory (TE) durations were altered, albeit minimally, in the animals with intact vagus nerves. Expiratory duration following hypoxia was not different from baseline levels either in vagotomized (P = 0.18) or intact (P > 0.05) animals. In contrast, injections of muscimol at more dorsal sites did not alter the decrease in frequency normally seen following hypoxia. 5. Histological examination revealed that effective lesion or injection sites were within the lateral pontine tegmental field and included portions of the noradrenergic A5 cell group. 6. We conclude that the mechanism responsible for post-hypoxic frequency decline involves an active neural process that depends on the integrity of the ventrolateral pons.