Over the past 300 years, the migraine field has been dominated by two main theories-the vascular theory and the central neuronal theory. The success of vasoconstrictors such as ergotamine and the triptans in treating acute migraine bolstered the vascular theory, but evidence is now emerging that vasodilatation is neither necessary nor sufficient to induce a migraine attack. Attention is now turning to the core migraine circuits in the brain, which include the trigeminal ganglia, trigeminal nucleus, medullary modulatory regions, pons, periaqueductal gray matter, hypothalamus and thalamus. Migraine triggers are likely to reflect a disturbance in overall balance of the circuits involved in the modulation of sensory activity, particularly those with relevance to the head. In this Review, we consider the evidence pointing towards a neuronal mechanism in migraine development, highlighting the role of calcitonin gene-related peptide (CGRP), which is found in small to medium-sized neurons in the trigeminal ganglion. CGRP is released during migraine attacks and can trigger migraine in patients, and CGRP receptor antagonists can abort migraine. We also examine whether other drugs, such as triptans, might exert their antimigraine effects via their actions on the neuronal circuit as opposed to the intracranial vasculature.