1. Intracellular recordings have been made from tonic and phasic expiratory, inspiratory and spiracular motoneurons and presumed sensory integrating neurons of the metathoracic ganglion in a locust during rhythmic ventilatory movements of the abdomen. The neurons have somata with diameters of no more than 30 <latex>$\mu$</latex>m situated on the ventral surface of the ganglion. 2. No motoneurons showed an intrinsic rhythmicity, all being driven in the ventilatory rhythm by complex patterns of synaptic inputs in one of the following ways: (a) excitation alone during the phase when spikes are produced (spiracle closer and some tonic expiratory motoneurons); (b) excitation during one phase and inhibition during the other (some tonic expiratory motoneurons); (c) excitation and inhibition during both phases (most motoneurons) in which one type of input dominates a particular phase. 3. The burst of spikes by a particular motoneuron may end because of a lack of excitatory input (spiracle closer motoneurons) or be terminated rapidly by inhibition (inspiratory motoneurons). Inhibition may also precede the main burst of spikes (inspiratory motoneurons) so that any spikes during the opposite phase are abolished. The pattern of synaptic input determines the frequency code of spikes within a burst. 4. Phasic expiratory motoneurons receive an underlying pattern of synaptic inputs in phase with ventilation even when they do not spike. Non-specific excitation (for example, a d.c. depolarization of the soma) is able to produce bursts of spikes in the correct phase of ventilation. 5. No direct pathway between any groups of motoneurons was found. Driving by common antecedent interneurons is inferred for those motoneurons which show similar patterns of spikes (inspiratory and spiracle closer motoneurons). 6. Stimulation of descending fibres in the pro-mesothoracic connectives evokes e.p.s.ps in some motoneurons, perhaps monosynaptically. In inspiratory motoneurons these fibres cause e.p.s.ps but will abolish the inspiratory burst of spikes and reset the ventilatory rhythm. All observations imply that the mechanism of the ventilatory rhythm lies among interneurons.