This article reviews experimental and clinical data on the use of magnesium as a neuroprotective agent in various conditions of cerebral ischemia. Whereas magnesium has shown neuroprotective properties in animal models of global and focal cerebral ischemia, this effect could not be reproduced in a large human stroke trial. These conflicting results may be explained by the timing of treatment. While treatment can be started before or early after ischemia in experimental studies, there is an inevitable delay of treatment in human stroke. Magnesium administration to women at risk for preterm birth has been investigated in several randomized controlled trials and was found to reduce the risk of neurological deficits for the premature infant. Postnatal administration of magnesium to babies after perinatal asphyxia has been studied in a number of controlled clinical trials. The results are promising but the trials have, so far, been underpowered. In aneurysmal subarachnoid hemorrhage (SAH), cerebral ischemia arises with the onset of delayed cerebral vasospasm several days after aneurysm rupture. Similar to perinatal asphyxia in impending preterm delivery, treatment can be started prior to ischemia. The results of clinical trials are conflicting. Several clinical trials did not show an additive effect of magnesium with nimodipine, another calcium antagonist which is routinely administered to SAH patients in many centers. Other trials found a protective effect after magnesium therapy. Thus, it may still be a promising substance in the treatment of secondary cerebral ischemia after aneurysmal SAH. Future prospects of magnesium therapy are discussed.