1. The three different types of synchronization of intercostal motoneurones which were described in the preceding paper (Kirkwood, Sears, Tuck & Westgaard, 1982) were studied for motoneurones of the same segment or for different segmental separations (up to five segments) and for motoneurones on opposite sides of the same segment.2. The strength of synchronization declined with segmental separation for all three categories, although the rate of decline was more variable for broad-peak synchronization than for the two other types. Short-term synchronization was undetectable for separations greater than three or four segments but clear peaks were still visible in the cross correlation histograms for high-frequency oscillation (h.f.o.) or broad-peak synchronization at a segmental separation of five. Synchronization between motoneurones on opposite sides of the cord was generally weak although less so for broad-peak or h.f.o. components.3. The decline in strength of short-term synchronization with segmental separation could not be explained by temporal dispersion of impulses in presynaptic axons.4. A time shift was observed in the position of the cross-correlation histogram peak which was dependent on segmental separation and equivalent to a mean descending conduction velocity of 28 m/s in the assumed common input. This figure is similar to the mean conduction velocity of bulbospinal respiratory neurones derived from published values.5. We conclude that the short-term synchronization in these preparations is generated by the bulbospinal respiratory neurones and that the majority of their axons do not branch to make strong synaptic connexions to motoneurones over more than three to four segments.6. Interpretations of the different distributions of the other types of synchronization are discussed.