Recent debate concerning the interpretation of studies of regional cerebral blood flow in migraine has re-emphasized that cerebral ischaemia may occur during attacks of migraine with aura. In this article we suggest that the presence of ischaemia during attacks makes it possible that migraine with aura causes neuronal damage in the long term. We argue that damage is likely to occur in the primary visual cortex, given that a recent high-resolution rCBF study has found flow reductions confined to this area. Furthermore, we hypothesize that the extent to which rCBF is reduced in migraine with aura is sufficient to cause damage only to GABA-ergic inhibitory interneurons in layer IV of this cortex. In animal models, similar cells are known to be selectively vulnerable to damage as a result of hypoxic conditions. Evidence consistent with our hypothesis is provided by recent studies of visual function in migraine. Some clinical and pathophysiological implications of this hypothesis are discussed.