In order to determine the anatomical distribution of cells concerned with relaying craniovascular nociception, local cerebral glucose utilization was determined by the 2-deoxyglucose method in tissue autoradiographs of the alpha-chloralose anesthetized cat. The superior sagittal sinus was carefully lifted from the brain by sectioning the dura laterally and the falx inferiorly and suspending the sinus across two platinum hook electrodes for stimulation. The sinus was stimulated electrically and its effect on caudal brainstem, upper cervical spinal cord and diencephalic metabolic activity determined. Stimulation of the sinus caused increased metabolic activity in the trigeminal nucleus caudalis, in the cervical dorsal horn and in a discrete area in the dorsolateral spinal cord at the second cervical segment. Metabolic activity was also increased in the ventrobasal thalamus, specifically in the ventroposteromedial (188%) nuclear group, in the medial nucleus of the posterior complex (70%) and the intralaminar complex (49%). There was no change in the surrounding thalamus, lateral geniculate nucleus or overlying cerebral cortex. These increases in 2-deoxyglucose utilisation were blocked by bilateral trigeminal ganglion ablation. The dorsolateral area activated in the spinal cord corresponds to a hitherto unrecognised group of cells in or near the lateral cervical nucleus that may form an important relay for craniovascular nociception. Further electrophysiological studies with glass coated tungsten microelectrodes have characterised the cells in these regions of the thalamus to be responsible for relaying nociceptive information. An understanding of the connections and properties of the neurons that subserve craniovascular pain is an essential prerequisite to understanding the complex pathophysiology of migraine.