1. The modulation of reflex responses in up to 10 simultaneously recorded hindlimb muscles was studied in unrestrained cats walking on a treadmill. Single electrical shocks of various strengths were applied to different skin areas of teh hindlimb at different times of the step cycle while the resulting EMG responses were sampled and analyzed. 2. Two excitatory response peaks (P1 and P2) at a latency of about 10 and 25 ms, respectively, were seen in all flexors examined (sartorius, semitendinosus, tibialis anterior, extensor digitorum longus). Stimulation of most skin areas was effective but responses were most easily obtained from stimuli applied to the foot or ankle. During the step cycle there was a marked modulation of the amplitudes of the responses, especially the P2 responses, which grew larger toward the end of stance when a maximum was reached, followed by a steady decline throughout swing. This pattern was very similar for various flexors, although these muscles differed considerably in their normal EMG activity pattern during walking. 3. Flexor responses were absent when the same stimuli were applied during the early stance phase. Instead, inhibition of the ongoing EMG activity was seen at a latency of 10 ms or less in all extensors examined (semimembranosus, quadriceps, soleus, gastrocnemius medialis, flexor digitorum longus). The inhibition was followed by a late excitatory peak (P3) at about 35-ms latency in all extensors except soleus. 4. Certain stimulation sites yielded exceptions to the above patterns. Stimulation of the skin area innervated by the sural nerve yielded larger and earlier MG excitatory responses as compared to stimulation of other skin areas. Activation of the plantar surface of the foot often failed to elicit P2 responses in the hip flexor sartorius, which showed inhibition instead. 5. In the hindlimb contralateral to the stimulus, excitatory responses occurred both in flexors and extensors at a latency of 20-25 ms. The pattern of modulation of these responses was similar to the ipsilateral modulation of P2 flexor and P3 extensor responses. Soleus failed to show a crossed response. 6. The data indicate that flexor and extensor responses differ both with respect to their latency and to their correlation with the ongoing EMG reactivity. It is concluded that these stimuli do not demonstrate reflex reversal in the strict sense in the normal walking cat but that there is modulation of transmission in a flexor excitatory and extensor inhibitory pathway, possibly by the flexor part of the spinal locomotor oscillator. In addition, there are some specialized flexor inhibitory and extensor excitatory pathways. The slow soleus muscle does not seem to be excited through these pathways.