We have used electrophysiological techniques to document segmental differences in the pathways between the giant, escape command axons, lateral giants (LG) and medial giants (MG), and the nongiant, fast flexor (FF) motoneurons. We found no difference in the input from LG and MG axons to FF motoneurons in the posterior (4th and 5th) ganglia. Since flexor motor output in these segments would be inconsistent with the LG-evoked behavior pattern, this finding was puzzling. Electromyographic (EMG) recordings during escape responses by intact unrestrained animals confirm that the FF muscles innervated by the posterior ganglia are not excited during LG-mediated tailflips, but are excited during MG-mediated tailflips. In the 2nd and 3rd ganglia, the command axons fire the FF motoneurons with high probability, in part via electrical excitatory postsynaptic potentials (EPSPs) from premotor neurons, the segmental giants (SG). In the 4th and 5th ganglia, the equivalent pathway is much less effective. Single, directly elicited impulses in SGs in ganglia 2 and 3 fire their respective FF motoneurons with high probability, while those in ganglia 4 and 5 rarely fire FF motoneurons. The command axons fire the SGs reliably in all segments. The amplitude of the SG-evoked EPSP in FF motoneurons is significantly smaller in posterior vs. anterior ganglia. For technical reasons, we are unable to present conclusive evidence on ganglionic variations in FF-motoneuron thresholds. The FF motoneurons receive additional excitatory input from intersegmental interneurons recruited by the command neurons. Motoneurons in ganglia 4 and 5 are excited by large interneurons that do not synapse on motoneurons in ganglia 2 and 3, but this additional input is not sufficient to compensate for the weaker effect of SG input. Unlike the all-or-none segmental differences demonstrated previously for the LG-to-motor giant pathway (24), the SG-to-FF pathway changes gradually, retains significant though subthreshold strength in posterior ganglia, and is common to both LGs and MGs. These features provide opportunities for variation in the spatial patterning of flexion and in the resulting escape trajectories.