In lactating rats, the milk ejection reflex is evoked and maintained by stimulation of the nipples by the suckling young. In order to understand the processing of the suckling stimulus within the spinal cord, urethane-anaesthetized lactating rats were prepared for electrophysiological recording from the thoraco-lumbar spinal cord during stimulation of the nipples. Single shocks to inguinal or abdominal nipples evoked a cord dorsum potential, consisting of an early (2.6 ms) afferent volley followed by a negative wave (100–200 μV, latency 5–7 ms, duration 5–10 ms). Evoked potentials were also recorded at various depths within the spinal cord, with a maximum amplitude (200–400 μV) at a depth of 400–800 μm, 400–800 μm lateral to the mid-line. At a given recording site, the response was maximal for one particular nipple but submaximal potentials could be evoked from adjacent nipples. Simultaneous stimulation of adjacent nipples caused summation of the response. Unit recordings were made from 35 spinal neurons. Upon electrical stimulation of the nipples, the cells responded with an early train of spikes (latency 5–15 ms) and in 6 cells, a later response (140–180 ms), with a higher stimulation threshold, was also observed. All cells examined showed convergence and summation from different nipples. Twenty out of 27 cells were also activated by stretching of the nipples, which evoked a rapidly adapting response; rhythmical stretching produced a more sustained increase in activity. The cells also responded to other natural stimuli such as touch and pressure or stroking the hair around the nipple and had large receptive fields. Six cells were tested with the suckling stimulus. There was a brisk increase in firing as the pup grasped the nipple and then intermittent (every 18–30 s) episodes of enhanced activity, which directly correlated with the suckling movements. These episodes continued for the duration of the suckling test and were enhanced when a second pup was placed on an adjacent nipple. Finally, from a few experiments when a stimulating electrode was placed within the contralateral antero-lateral funiculus at the level of C2–C3 for antidromic identification, it was seen that some of the cells activated from the nipples projected to higher levels. The short latency responses to nipple stimulation, including suckling, suggest that the suckling stimulus reaches the spinal cord ungated. On the other hand, the results showing convergence and summation of the afferent input on spinal neurons demonstrate that some integration of the suckling stimulus already occurs within the spinal cord before further processing takes place at higher levels. The possible significance of these observations in relation to the control of the milk ejection reflex is discussed.