1. The isolated brainstem of larval Rana catesbeiana maintained in vitro generates neural bursts that correspond to the lung and gill ventilatory activity generated in the intact specimen. To investigate the role of chloride channel-dependent inhibitory mechanisms mediated by GABA(A) and/or glycine receptors on fictive lung and gill ventilation, we superfused the isolated brainstems with agonists, antagonists (bicuculline and/or strychnine) or a chloride-free solution while recording multi-unit activity from the facial motor nucleus. 2. Superfusion with the agonists (GABA or glycine) produced differential effects on frequency, amplitude and duration of the neural bursts related to lung and gill ventilation. At a GABA or glycine concentration of 1.0 mM, fictive gill bursts were abolished while fictive lung bursts persisted, albeit with reduced amplitude and frequency. 3. At the lowest concentrations used (1.0-2.5 microM), the GABA(A) receptor antagonist bicuculline produced an increase in the frequency of lung bursts. At higher concentrations (5.0-2.0 microM) bicuculline produced non-specific excitatory effects. The glycine antagonist strychnine, at concentrations lower than 5.0 microM, caused a progressive decrease in the frequency and amplitude of the gill bursts and eventually abolished the rhythmic activity. At higher concentrations (7.5 microM), non-specific excitatory effects occurred. Superfusion with bicuculline (10 microM) and strychnine (5 microM) combined abolished the neural output for gill ventilation but increased the frequency, amplitude and duration of lung bursts. 4. Superfusion with Cl(-)-free solution also abolished the rhythmic neural bursts associated with gill ventilation, while it significantly increased the amplitude (228 +/- 51%; P < 0.05) (mean +/- S.E.M.) and duration of the lung bursts (3.5 +/- 0.1 to 35.3 +/- 3.7 s; P < 0.05) and improved the regularity of their occurrence. 5. We conclude that different neural systems generate rhythmic activity for lung and gill ventilation. Chloride-mediated inhibition may be essential for generation of neural bursts associated with gill ventilation. In contrast, the burst associated with lung ventilation can be generated in the absence of Cl(-)-mediated inhibition although the latter plays a role in shaping the normal lung burst.