To study the potential plasticity of locomotor networks in the spinal cord, an important issue for locomotor rehabilitation after spinal injuries, we have investigated the locomotor performance of cats before and after a unilateral denervation of the ankle flexors tibialis anterior (TA) and extensor digitorum longus (EDL) both in cats with intact spinal cord and after spinalization. The effects of the inactivation of the ankle flexors were studied in three cats with intact spinal cord during periods of 4-7 wk. Cats adapted their locomotor performance very rapidly within a few days so that the locomotor behavior appeared to be unchanged practically. However, kinematic analyses of video records often revealed small but consistent increase in knee and/or hip flexion. These changes were accompanied by some increase in the amplitude of knee and hip flexor muscle activity. Cats maintained a regular and symmetrical walking pattern over the treadmill for several minutes. Two of these cats then were spinalized at T13 and studied for approximately 1 mo afterward. Whereas normally cats regain a regular and symmetrical locomotor pattern after spinalization, these cats had a disorganized and asymmetrical locomotor pattern with a predominance of knee flexion and absence of plantar foot contact of the denervated limb. Another cat first was spinalized and allowed to recuperate a regular symmetrical locomotor performance. Then it also was submitted to the same unilateral ankle flexor inactivation and studied for approximately 50 days. The cat maintained a well-organized symmetrical gait although there was almost no ankle flexion on the denervated side. There was no exaggerated knee hyperflexion and gait asymmetry as seen in the two previous cats spinalized only after they had adapted to the denervation of ankle flexors. It is concluded that, after muscle denervation, locomotor adaptation is achieved through changes occurring at different levels. Because cats spinalized after adaptation to the neurectomy had an asymmetrical locomotor pattern dominated by hyperflexion, it is suggested that the spinal circuitry has been modified during the adaptive process, presumably through the action of corrective supraspinal inputs. Indeed spinal cats do not normally display such abnormal hyperflexions, and neither did the one cat denervated after spinalization. On the other hand, because the modified locomotor pattern in the spinal state is not functional and contains only some aspects of the compensatory response seen before spinalization, it is suggested that the complete functional adaptation observed in intact cats after peripheral nerve lesions may depend on changes occurring at the spinal and the supraspinal levels.