The underlying principle in interpreting data from positron emission tomography (PET) is that increases and decreases of synaptic activity in the brain are accompanied by equivalent changes in regional blood flow. However, given the relatively low spatial resolution of PET, regional indices of neural activity can be influenced by signals from nonneural elements in the intracranial space. We have used PET to study the functional neuroanatomy of headache syndromes, such as cluster headache and experimental headache, by observing activation in brain that is due to synaptic activity. In these studies we noted a bilateral activation in midline structures over several planes that, based on its anatomy, is likely to arise from bilateral large intracranial arteries in the region of the cavernous sinus. We have observed this activity in two different group studies and in 11 of 17 single-subject studies. The most likely explanation for this "activation" is the increased volume of space occupied by dilated large vessels containing H215O. We further defined the anatomy of the PET findings using magnetic resonance angiography, which showed bilateral vasodilatation of the internal carotid artery and the basilar artery in nitroglycerin-induced cluster headache attacks. While the fact that vascular structures can contaminate PET-based blood flow studies may be well known to methodological experts, the issue is less well recognized outside the immediate field.