1. This study examines rhythmical activity of primary afferents occurring during "fictive" locomotion in decorticate paralyzed cats. Oscillations of the dorsal root potential (DRP) at the frequency of the locomotor rhythm have been observed at the lumbosacral and cervical levels. In addition, rhythmic antidromic discharges of primary afferent units have been recorded from the proximal stumps of cut dorsal root filaments. A detailed study of the relationships between the DRP fluctuations, the antidromic discharges, and the locomotor activity monitored by recording extensor and flexor muscle nerves is presented. 2. Typical DRP recordings from both lumbosacral and cervical levels show two negative waves (N1 and N2) separated by positive troughs (P1 and P2) in each locomotor cycle. Linear regression analyses indicate that the first negative wave (which generally has the largest amplitude) is related to the flexor activity whereas the second is related to the extensor activity. The relative amplitude of the two negative waves may vary without apparent concomitant changes in the recorded flexor or extensor motor nerves. The positive troughs occur respectively close to the period of transition between flexor and extensor activities and between extensor and flexor activities. 3. DRPs of similar period and amplitude can be observed in different ipsilateral roots recorded simultaneously. The DRPs recorded bilaterally from the same segment have the same periodicity but are out-of-phase. Point-to-point variations of amplitude in bilaterally recorded roots are not correlated. This suggests that the polarization of primary afferents on one side is mainly related to the locomotor events on that side. DRPs have been recorded in cats spinalized at Th13 and injected with nialamide and l-DOPA. This suggests that although the supraspinal contribution may be important, at least part of the DRPs may result from locomotor activity within the spinal cord itself. 4. A salient finding in our experiments was that of rhythmic antidromic unit discharges in the proximal stump of cut dorsal root filaments. Of the 194 units recorded, 19% (37/194) discharged in distinct bursts occurring at fixed times in the locomotor cycle. The majority of the units discharged either one burst during the period of flexor or extensor activity or one burst during one of the two periods of transition. Three units discharged two bursts per locomotor cycle. The frequency of the antidromic discharges of some units in one limb were also found to be modulated by stimulation of the skin or passive manipulation of the limbs.(ABSTRACT TRUNCATED AT 400 WORDS)