1. When an isolated metathoracic ganglion of the locust was superfused with the muscarinic cholinergic agonist pilocarpine, rhythmic activity was induced in leg motor neurons. The frequency of this induced rhythm increased approximately linearly from 0 to 0.2 Hz with concentrations of pilocarpine from 10(-5) to 10(-4) M. Rhythmic activity evoked by pilocarpine could be completely and reversibly blocked by 3 x 10(-5) M atropine, but was unaffected by 10(-4) M d-tubocurarine. 2. For each hemiganglion, the observed rhythm was characterized by two main phases: a levator phase, during which the anterior coxal rotator, levators of the trochanter, flexors of the tibia, and common inhibitory motor neurons were active; and a depressor phase, during which depressors of the trochanter, extensors of the tibia, and depressors of the tarsus were active. Activity in depressors of the trochanter followed the activity of the levators of the trochanter with a short, constant interburst latency. Activity in the levator of the tarsus spanned both phases. 3. The levator phase was short compared with the period (0.5-2 s, or 10-20% of the period) and did not depend on the period. The interval between the end of a levator burst and the beginning of the following one thus increased with cycle period. The depressor phase was more variable, and was usually shorter than the interval between successive levator bursts. 4. Motor neurons in a same pool often received common discrete synaptic potentials (e.g., levators of trochanter or extensors of tibia), suggesting common drive during the rhythm. Coactive motor neurons on opposite sides (such as left trochanteral depressors and right trochanteral levators), however, did not share obvious common postsynaptic potentials. Depolarization of a pool of motor neurons during its phase of activity was generally accompanied by hyperpolarization of its antagonist(s) on the same side. 5. Rhythmic activity was generally evoked in both hemiganglia of the metathoracic ganglion, but the intrinsic frequencies of the rhythms on the left and right were usually different. The activity of the levators of the trochanter on one side, however, was strongly coupled to that of the depressors of the trochanter on the other side. 6. The locomotory rhythm was weakly coupled to the ventilatory rhythm such that trochanteral levator activity on either side never occurred during the phase of spiracle opener activity corresponding to inspiration. 7. The rhythmic activity observed in vitro bears many similarities to patterns of neural and myographic activity recorded during walking. The similarities and differences are discussed.