In lactating rats, suckling elicits the milk ejection reflex which consists of an intermittent synchronous activation of hypothalamic oxytocin neurons which releases oxytocin into the bloodstream. We here investigated the electrophysiological behaviour of spinal cord neurons linked to mammary innervation in relation to suckling and the suckling-induced milk ejection reflex. Experiments were carried out on 58 urethane-anaesthetized rats, paralysed with gallamine triethiodide and artificially ventilated. Fixation of the spinal cord and laminectomy significantly slowed down the reflex, which occurred in only 27% of the animals. In these rats, 31 dorsal horn neurons at the thoraco-lumbar level were found to be excited by nipple stimulation. During suckling by a litter of at least nine pups, they displayed an irregular pattern of brief bursts of activity (peak firing rate 22.0 ± 3.2 Hz, mean ± SD) correlated to the bouts of suckling of the pups. Seventeen out of 19 cells tested by stimulation of at least 2 adjacent nipples received convergent input from different ipsilateral nipples. Out of 11 cells tested, 8 were also activated by stimulation of a contralateral nipple. Fourteen out of 30 units were recorded through at least one reflex milk ejection. Their firing rate was significantly higher than the firing rate of cells recorded in animals which failed to milk eject (4.4 Hz ± 2.8 versus 1.5 Hz ± 0.7). At the moment of the high frequency discharge of action potentials, occurring in oxytocinergic cells 10 to 15 s before each milk ejection, spinal neurons showed no systematic change in electrical activity. In contrast, the stretch reaction of the pups, which corresponds to an intense period of suckling when milk ejection has started, induced, in 12 cells, a considerable increase in electrical activity. One unit was found to be inhibited by suckling and during the stretch reaction. Ten more units, which were not activated by stimulation of the nipples but responded to stimulation of excitatory receptive fields near the last three pairs of nipples, were recorded through reflex milk ejections: 8 remained silent during reflex milk ejections but 2 were activated when the pups stimulated their excitatory receptive field. We conclude that some dorsal horn neurons, able to respond readily to the suckling movements of pups, appear to receive an ungated input from the nipples. At the time of the activation of oxytocin neurons, they present no particular pattern of activation or inhibition which could account in a simple manner for the intermittence of the high frequency discharge in oxytocinergic cells. However, in so far as these dorsal horn neurons may be part of the milk ejection reflex pathway, their activity, showing convergence and summation of input, and being facilitated in milk ejecting animals, indicates that the reflex does undergo a certain degree of processing at the spinal cord level.