A technique was developed to measure, in the anesthetized and paralyzed cat under artificial ventilation, changes of excitability to intraspinal stimulation simultaneously in two different afferent fibers or in two collaterals of the same afferent fiber. Intraspinal stimulation reduced the threshold of single muscle afferent fibers ending in the intermediate nucleus. This effect was seen with strengths below those required to activate the afferent fiber tested (1.5-12 microA), occurred at a short latency (1.5-2.0 ms), reached a maximum between 15 and 30 ms, and lasted up to 100 ms. The effects produced by graded stimulation applied at the shortest conditioning-testing stimulus time intervals increased by fixed steps, suggesting recruitment of discrete elements, most likely of last-order interneurons mediating primary afferent depolarization (PAD). The short-latency increases in excitability produced by the weakest effective intraspinal stimuli were usually detected only in the collateral closest to the stimulating micropipette, indicating that the stimulated interneurons mediating PAD have spatially restricted actions. The short-latency PAD produced by intraspinal stimuli, as well as the PAD produced by stimulation of the posterior biceps and semitendinosus (PBSt) nerve or by stimulation of the bulbar reticular formation (RF), was depressed 19-30 min after the i.v. injection of 0.5 mg/kg of picrotoxin, suggesting that all these effects were mediated by GABAergic mechanisms. The PAD elicited by stimulation of muscle and/or cutaneous nerves was depressed following the i.v. injection of (-)-baclofen, whereas the PAD elicited in the same collateral by stimulation of the RF was baclofen-resistant. The short-latency PAD produced by intraspinal stimulation was not always depressed by i.v. injections of (-)-baclofen. Baclofen-sensitive and baclofen-resistant monosynaptic PADs could be produced in different collaterals of the same afferent fiber. The results suggest that the intraspinal terminals of single muscle afferents receive synapses from more than one PAD-mediating GABAergic interneuron and that a single last-order interneuron has synaptic connections with a restricted number of intraspinal terminals and/or collaterals of the same afferent fiber. In addition, they support the existence of separate subsets of last-order baclofen-sensitive and baclofen-resistant interneurons that respond predominantly to segmental and to descending inputs. It is suggested that the restricted nature of the PAD plays an important role in the central control of the synaptic effectiveness of group I muscle afferents.