A rapid plantar flexion perturbation applied to the ankle during the stance phase of the step cycle during human walking unloads the ankle extensors and produces a marked decline in the soleus EMG. This demonstrates that sensory activity contributes importantly to the enhancement of the ankle extensor muscle activation during human walking. On average, the EMG begins to decline approximately 52 ms after the perturbation. In contrast, a rapid dorsi flex ion perturbation produces a group Ia mediated short-latency stretch reflex burst with an onset latency of approximately 36 ms. The transmission of sensory traffic from the foot and ankle was suppressed in 10 subjects by an anaesthetic nerve block produced with local injections of lidocaine hydrochloride. The anaesthetic block had no effect on the stance phase soleus EMG, the latencies of the EMG responses, or the magnitude of the EMG decline following the plantar flexion perturbation. Therefore, it is more likely that proprioceptive afferents, rather than cutaneous afferents, contribute to the background soleus EMG during the late stance phase of the step cycle. The large difference in onset latencies between the short-latency reflex and unload responses suggests that the largest of the active group Ia afferents might not contribute strongly to the background soleus EMG, although it remains to be determined which of the proprioceptive pathways provide the more important contributions.Key words: afferent feedback, gait, locomotion, stretch reflex.