1. In ten decerebrate, paralysed and ventilated cats, we tested the hypothesis that cervical (C5) respiratory interneurones mediate inhibition of phrenic motoneurone activity resulting from single shocks to the phrenic nerve. 2. Stimulus intensities sufficient to activate all afferents elicited (latency, 4.0 +/- 0.9 ms, mean +/- S.D.) a graded suppression of ipsilateral, but not contralateral (five of seven cats) phrenic nerve activity lasting, in six of seven cats, more than 70 ms and interrupted by a brief (approximately 6-18 ms duration) excitation at latencies between 7 and 30 ms. 3. In twenty-five ipsilateral motoneurones, peristimulus time average of the membrane potentials (-61 +/- 10 mV) showed no effect in eleven; of the fourteen that responded, ten had initial EPSPs (latency, 17.6 +/- 3.0 ms) and four initial IPSPs (latencies, 2.25-4.3 ms). Only one motoneurone had both. No responses with latencies > 60 ms were observed. 4. Peristimulus time averages of extracellular activity of thirty ipsilateral interneurones, twenty-five firing in inspiration (I) and five in expiration (E), showed diverse responses. The initial response of I interneurones was an excitation in eleven, a suppression of activity in nine, and no response in five. Latencies of excitations ranged from 2 to 36.5 ms (median, 14 ms) with durations ranging from 2 to 7 ms (mean, 4.4 +/- 1.6 ms). Latencies of suppression of activity ranged from 2 to 29 ms (median, 10 ms). Two E interneurones were excited (latencies, 11 and 15 ms; durations, 3.5 and 2 ms), two inhibited (latencies, 2 and 12 ms; durations, > 40 and 17 ms, respectively), and one did not respond. 5. In nine interneurones (seven I, two E), peristimulus time averages of the membrane potentials (mean, -62 +/- 14 mV) revealed no effect on three (all I). Of the six that responded, four (three I) had initial IPSPs, two (one I, one E) initial EPSPs. EPSPs had latencies of 11.5 (I interneurone) and 22 ms (E interneurone); the latencies of the IPSPs were 2.75, 3.20, and 2.3 ms for the I interneurones and 15.9 ms for the E interneurone). No responses with latencies > 30 ms were observed. 6. The diverse responses of cervical respiratory interneurones indicates that they do not mediate the prolonged suppression of ipsilateral phrenic activity elicited by stimulation of phrenic afferents. The suppression may result from activation of normally quiescent inhibitory interneurones or from presynaptic inhibition.