1. An intracellular search was undertaken in an effort to locate and characterize the elements involved in command-driven positional behavior in the isolated abdominal nerve cord of the crayfish,Procambarus clarkii. Small bundles containing flexion or extension-evoking command fibers were isolated in the anterior connectives; these bundles are considered here to be equivalent to identified command fibers. A microelectrode filled with Lucifer Yellow CH was used to find elements within the fourth abdominal ganglion that showed synaptic responses to either flexion or extension command fibers; and these elements were further characterized by current and dye injections. 2. The interneurons thus encountered have been classified primarily by the form of motor output they evoked when strongly depolarized, and secondarily by their morphologies. We present data in this paper from 12 types of flexion-evoking cells, 4 types of extension-evoking cells and one inhibitory type that we term ‘flexion-antagonistic’. 3. With the exception of this latter group, many of the interneurons evoked a fully or nearly fully organized motor output. That is, flexion-evoking interneurons activated the flexor motoneurons and the peripheral inhibitor to the extensor muscles, and usually inhibited much of the activity in the extensor motoneurons. Conversely, the extension-evoking interneurons activated the extensor motoneurons and the peripheral inhibitor to the flexor muscles and usually inhibited the flexor motoneurons. The flexion-antagonists inhibited the flexor motoneurons and the peripheral inhibitor to the extensor muscles, but did not activate any motoneurons. 4. Because of the search strategy used, the majority of the cells that were studied showed synaptic connections to one or more of the three flexion and one extension command fiber bundle stimulated. Both apparently monosynaptic and polysynaptic connections were evident, but very few inhibitory connections were found. Thus, while flexion-evoking interneurons were in general depolarized by the flexion CFs, they were not hyperpolarized by the extension CF. Furthermore, we have demonstrated that several flexion-evoking interneurons share inputs from more than one flexion CF, whereas others have inputs from only one of three flexion CFs. Finally, at least two interneurons were found (type 5 and 6) that were apparently independent of the command neurons used in this study. 5. Six of the flexion-evoking types were seen to have their somata within ganglion 4 and all had axonal processes extending either anteriorly or posteriorly. The remaining 6 flexor interneurons were seen only as an axonal process running through ganglion 4. We cannot determine from the present data their origins or the extent of their axons in the nerve cord. In several preparations, both forms of flexion-evoking cells were seen to produce motor output in the ganglia adjacent to G4, suggesting that these cells may be either command neurons or driver interneurons. Different experimental approaches will be required to resolve this question. 6. While we cannot at this point propose the organization of CF-driven postural behavior, we have encountered several potentially important elements involved in the control of abdominal flexion. The motor output appears to be achieved by a combination of direct synaptic connections from the CF to the motoneurons and through connections with several other interneurons acting in parallel. A combination of these connections may provide for the strong reciprocity between flexor and extensor motoneurons, since individual driver cells are not always capable of this reciprocal output. Finally, redundancy of driver cells is indicated since we have not been able to identify any critical elements that, when functionally removed by hyperpolarization, eliminate a significant portion of the CF driven motor output.