We have recently demonstrated that the brainstem respiratory network can operate in 3 modes, each reflecting a distinct spatial organization and rhythmogenic mechanism (Smith et al., 2007). Here, we focused on the possible mechanisms and the brainstem regions involved in the generation of expiratory activity under different conditions. For that, Wistar rats (males, 65-85g, n=12) were anaesthetised deeply with halothane until they became unresponsive to noxious pinching of the tail. They were decerebrated at the pre-collicular level and perfused arterially (Paton, 1996). The following approaches were used: (1) simultaneous recording of phrenic nerve (PN), central vagus nerve (cVN) and different levels of abdominal (AB) motor outflow (T₁₀-L₁), during rest and hypercapnia; (2) brainstem transections and chemical suppression of the retrotrapezoid nuclei (RTN) & ventrolateral parafacial region (vlPF). Values are mean ± SEM and were compared using t test. In the in situ decerebrate preparation of the rat with intact brainstem and pons, generating a 3-phase rhythm, AB expiratory activity was represented by a low amplitude post-inspiratory (post-I) discharge in all spinal levels recorded. However, during hypercapnia (7-10% CO₂) augmenting late-expiratory (late-E) discharges were evoked. Caudal branches of AB exhibited higher ratio of late-E/post-I activity (P<0.05, n=5). Transverse sectioning through the rostral pons revealed a 2-phase rhythm which exhibited tonic AB discharge throughout expiration and absence of post-I activity in the cVN (n=7). When the Bötzinger complex (BötC) was removed all expiratory activity was abolished and could not be reinstated by hypercapnia and/or hypoxia. Suppression of the RTN/vlPF (isoguvacine microinjections, 60 nL, 20 mM) attenuated resting post-I activity from AB (-69±14 %, n=5, P<0.05) and cVN (-68±12 %, n=5, P<0.05), which partially recovered during hypercapnia. Suppression of the RTN/vlPF region abolished late-E AB bursts evoked by hypercapnia. The most potent inhibition of expiratory motor outflow as obtained from microinjections in the caudal half of RTN/vlPF. We conclude that the integrity of RTN/vlPF is required for regeneration late-E AB bursts during hypercapnia. The BötC is essential for generation of expiratory activity on a 2-phase rhythm. The pons provides essential tonic excitatory drive to post-I cell populations in the brainstem with RTN/vlPF contributing to this drive.