Despite considerable research into the pathogenesis of idiopathic headaches, such as migraine, the pathophysiological mechanisms underlying them remain poorly understood. Although it is well established that the trigeminal nerve becomes activated during migraine, the consequences of this activation remain controversial. One theory, based on preclinical observations, is that activation of trigeminal sensory fibers leads to a painful neurogenic inflammation within the meningeal (dural) vasculature mediated by neuropeptide release from trigeminal sensory fibres and characterized by plasma protein extravasation, vasodilation, and mast cell degranulation. Effective antimigraine agents such as ergots, triptans, opioids, and valproate inhibit preclinical neurogenic dural extravasation, suggesting that this activity may be a predictor of potential clinical efficacy of novel agents. However, several clinical trials with other agents that inhibit this process preclinically have failed to show efficacy in the acute treatment of migraine in man. Alternatively, it has been proposed that painful neurogenic vasodilation of meningeal blood vessels could be a key component of the inflammatory process during migraine headache. This view is supported by the observation that jugular plasma levels of the potent vasodilator, calcitonin gene-related peptide (CGRP) are elevated during the headache and normalized by successful sumatriptan treatment. Preclinically, activation of trigeminal sensory fibers evokes a CGRP-mediated neurogenic dural vasodilation, which is blocked by dihydroergotamine, triptans, and opioids but unaffected by NK1 receptor antagonists that failed in clinical trials. These observations suggest that CGRP release with associated neurogenic dural vasodilation may be important in the generation of migraine pain, a theory that would ultimately be tested by the clinical testing of a CGRP receptor antagonist.