This study was primarily aimed at investigating the selectivity of the cortico-spinal actions exerted on the pathways mediating primary afferent depolarization (PAD) of muscle spindle and tendon organ afferents ending within the intermediate nucleus at the L6-L7 segmental level. To this end we analyzed, in the anesthetized cat, the effects produced by electrical stimulation of sensory nerves and of the cerebral cortex on (a) the intraspinal threshold of pairs of single group I afferent fibers belonging to the same or to different hindlimb muscles and (b) the intraspinal threshold of two collaterals of the same muscle afferent fiber. Afferent fibers were classified in three categories, according to the effects produced by stimulation of segmental nerves and of the cerebral cortex. Twenty-five of 40 fibers (62.5%) were depolarized by stimulation of group I posterior biceps and semitendinosus (PBSt) or tibialis (Tib) fibers, but not by stimulation of the cerebral cortex or of cutaneous and joint nerves, which instead inhibited the PBSt- or Tib-induced PAD (type A PAD pattern, usually seen in Ia fibers). The remaining 15 fibers (37.5%) were all depolarized by stimulation of the PBSt or Tib nerves and the cerebral cortex. Stimulation of cutaneous and joint nerves produced PAD in 10 of those 15 fibers (type B PAD pattern) and inhibited the PBSt- or Tib-induced PAD in the 5 remaining fibers (type C PAD pattern). Fibers with a type B or C PAD pattern are likely to be Ib. Not all sites in the cerebral cortex inhibited with the same effectiveness the segmentally induced PAD of group I fibers with a type A PAD pattern. With the weakest stimulation of the cortical surface, the most effective sites that inhibited the PAD of individual fibers were surrounded by less effective sites, scattered all along the motor cortex (area 4gamma and 6) and sensory cortex (areas 3, 2 and 1), far beyond the area of projection of group I fibers from the hindlimb. With higher strengths of cortical stimulation, the magnitude of the inhibition was also increased, and previously ineffective or weakly effective sites became more effective. Maps obtained when using the weakest cortical stimuli have indicated that the most effective regions that produced PAD of group I fibers with a type B or type C PAD pattern were also scattered throughout the sensory-motor cortex, in the same general area as those that inhibited the PAD of group I afferents with a type A PAD pattern. In eight fibers with a type A PAD pattern it was possible to examine the intraspinal threshold of two collaterals of the same single afferent fiber ending within the intermediate nucleus at the L7 segmental level. In six fibers, stimulation of the PBSt nerve with trains of pulses between 1.5 and 1.86 times threshold (xT) produced a larger PAD in one collateral than in the other. In seven fibers, stimulation of the sensory-motor cortex and of cutaneous nerves produced a larger inhibition of the PBSt-induced PAD in one collateral than in the other. The ratio of the cortically induced inhibition of the PAD elicited in the two collaterals could be modified by changing the strength of cortical and of PBSt stimulation. In three fibers it was possible to inhibit almost completely the background PAD elicited in one collateral while having little or no effect on the PAD in the other collateral. Changes in the intraspinal threshold of pairs of collaterals following electrical stimulation of segmental nerves and of the somato-sensory cortex were examined in three fibers with a type B and two fibers with a type C PAD pattern. In four fibers the PAD elicited by stimulation of cutaneous (4-20xT) and muscle nerves (1.54-3.7xT), or by stimulation of the sensory-motor cortex, was of different magnitude in the two collaterals. In two experiments it was possible to find cortical sites in which weak surface stimulation produced PAD in one collateral only. (ABSTRACT TRUNCATED)