It has been previously shown that phasic stimulation of group I afferents from ankle and knee extensor muscles may entrain and/or reset the intrinsic locomotor rhythm; these afferents are thus acting on motoneurones through the spinal rhythm generators. It was also concluded that the major part of these effects originates from Golgi tendon organ Ib afferents. Transmission in this pathway to lumbar motoneurones has now been investigated during fictive locomotion in spinal cats injected with nialamide and L-DOPA, and in decerebrate cats with stimulation of the mesencephalic locomotor region. In spinal cats injected with nialamide and L-DOPA, it was possible to evoke long-latency, long-lasting reflexes upon stimulation of high threshold afferents before spontaneous fictive locomotion commenced. During that period, stimulation of ankle and knee extensor group I afferents evoked oligosynaptic excitation of extensor motoneurones, rather than the "classical" Ib inhibition. Furthermore, a premotoneuronal convergence (spatial facilitation) between this group I excitation and the crossed extensor reflex was established. During fictive locomotion, in both preparations, the transmissions in these groups I pathway was phasically modulated within the step cycle. During the flexor phase, the group I input cut the depolarised (active) phase in flexor motoneurones and evoked EPSPs in extensor motoneurones; during the extensor phase the group I input evoked smaller EPSPs in extensor motoneurones and had virtually no effect on flexor motoneurones. The above results suggest that the group I input from extensor muscles is transmitted through the spinal rhythm generator and more particularly, through the extensor "half-centre". The locomotor-related group I excitation had a central latency of 3.5-4.0 ms. The excitation from ankle extensors to ankle extensors remained after a spinal transection at the caudal part of L6 segment; the interneurones must therefore be located in the L7 and S1 spinal segments. Candidate interneurones for mediating these actions were recorded extracellularly in lamina VII of the 7th lumbar segment. Responses to different peripheral nerve stimulation (high threshold afferents and group I afferents bilaterally) were in concordance with the convergence studies in motoneurones. The interneurones were rhythmically active in the appropriate phases of the fictive locomotor cycle, as predicted by their response patterns. The synaptic input to, and the projection of these candidate interneurones must be fully identified before their possible role as components of the spinal locomotor network can be evaluated.