1. Reflex responses were elicited in muscles that act at the ankle by electrical stimulation of low-threshold afferents from the foot in human subjects who were reclining supine. During steady voluntary contractions, stimulus trains (5 pulses at 300 Hz) were delivered at two intensities to the sural nerve (1.2-4.0 times sensory threshold) or to the posterior tibial nerve (1.1-3.0 times motor threshold for the intrinsic muscles of the foot). Electromyographic (EMG) recordings were made from tibialis anterior (TA), peroneus longus (PL), soleus (SOL), medial gastrocnemius (MG), and lateral gastrocnemius (LG) muscles by the use of intramuscular wire electrodes. 2. As assessed by averages of rectified EMG, stimulation of the sural or posterior tibial nerves at nonpainful levels evoked a complex oscillation with onset latencies as early as 40 ms and lasting up to 200 ms in each muscle. The most common initial responses in TA were a decrease in EMG activity at an onset latency of 54 ms for sural stimuli, and an increase at an onset latency of 49 ms for posterior tibial stimuli. The response of PL to stimulation of the two nerves began with a strong facilitation of 44 ms (sural) and 49 ms (posterior tibial). With SOL, stimulation of both nerves produced early inhibition beginning at 45 and 50 ms, respectively. With both LG and MG, sural stimuli produced an early facilitation at 52-53 ms. However, posterior tibial stimuli produced different initial responses in these two muscles: facilitation in LG at 50 ms and inhibition in MG at 51 ms. 3. Perstimulus time histograms of the discharge of 61 single motor units revealed generally similar reflex responses as in multiunit EMG. However, different reflex components were not equally apparent in the responses of different single motor units: an individual motor unit could respond slightly differently with a change in stimulus intensity or background contraction level. The multiunit EMG record represents a global average that does not necessarily depict the precise pattern of all motor units contributing to the average. 4. When subjects stood erect without support and with eyes closed, reflex patterns were seen only in active muscles, and the patterns were similar to those in the reclining posture. 5. It is concluded that afferents from mechanoreceptors in the sole of the foot have multisynaptic reflex connections with the motoneuron pools innervating the muscles that act at the ankle. When the muscles are active in standing or walking, cutaneous feedback may play a role in modulating motoneuron output and thereby contribute to stabilization of stance and gait.