Migraine headache is widely believed to be associated with cerebral or meningeal vasodilatation. Human evidence for this hypothesis is lacking. 3 Tesla magnetic resonance angiography (3T MRA) allows for repetitive, non-invasive, sensitive assessment of intracranial vasodilatation and blood flow. Nitroglycerine (NTG) can faithfully induce migraine attacks facilitating pathophysiological studies in migraine. Migraineurs (n = 32) randomly received NTG (IV 0.5 microg/kg/min for 20 min; n = 27) or placebo (n = 5; for blinding reasons). Using 3T MRA, we measured: (i) blood flow in the basilar (BA) and internal carotid arteries (ICA) and (ii) diameters of the middle meningeal, external carotid, ICA, middle cerebral, BA and posterior cerebral arteries at three timepoints: (a) at baseline, outside an attack; (b) during infusion of NTG or placebo and (c) during a provoked attack or, if no attack had occurred, at 6 h after infusion. Migraine headache was provoked in 20/27 (74%) migraineurs who received NTG, but in none of the five patients who received placebo. The headache occurred between 1.5 h and 5.5 h after infusion and was unilateral in 18/20 (90%) responders. During NTG (but not placebo) infusion, there was a transient 6.7-30.3% vasodilatation (P < 0.01) of all blood vessels. During migraine, blood vessel diameters were no different from baseline, nor between headache and non-headache sides. There were no changes in BA and ICA blood flow during either NTG infusion or migraine. In contrast to widespread belief, migraine attacks are not associated with vasodilatation of cerebral or meningeal blood vessels. Future anti-migraine drugs may not require vasoconstrictor action.