Intensely noxious peripheral stimuli of the anaesthetized rat produce two changes in the activity of convergent dorsal horn units: the segmental neuronal pool is activated, while all other convergent neurones are inhibited. These Diffuse Noxious Inhibitory Controls (DNIC) are highly potent (60-80% inhibition) and suppress all convergent neuronal activity, whether spontaneous or evoked by noxious or nonnoxious stimuli. On the other hand, they have no effect on other dorsal horn cell types, including noxious-only and proprioceptive units. The" DNIC" circuits include at least one supraspinal relay since DNIC is not seen in spinal animals. Furthermore, they are greatly reduced by lesions of the Nucleus Raphé Magnus (NRM). It has been shown that this nucleus massively projects onto the spinal cord, in particular onto the dorsal horn, and that stimulation of the NRM results in convergent unit inhibition of the same degree of magnitude as with DNIC. The role of serotonergic mechanisms in DNIC can be demonstrated pharmacologically: pCPA pre-treatment (3 daily IP injections, 300 mg/kg) or cinanserin (4 mg/kg IV) both result in a potent decrease (50-80%). We have proposed that the nociceptive message from the convergent units could result in a contrast between activity of the activated segmental pool and silence of the remaining convergent units. If this hypothesis can be verified, then some raphé nuclei and brain stem serotonergic pathways may function as filters in the detection of nociceptive messages, allowing extraction of information from somatic background activity including the firing from peripheral mechanoreceptors. While superficially paradoxical in fact our hypothesis fits well with the observation of profound analgesia following NRM stimulation: indeed, this hypothetical contrast would be completely eliminated by NRM stimulation since both neuronal pools would then be inhibited.