Ever since the hypothesis was put forward that estrogens could protect against cerebral ischemia, numerous studies have investigated the mechanisms of their effects. Despite initial studies showing ameliorating effects, later trials in both humans and animals have yielded contrasting results regarding the fundamental issue of whether estrogens are neuroprotective or neurodamaging. Therefore, investigations of the possible mechanisms of estrogen actions in brain ischemia have been difficult to assess. A recently published systematic review from our laboratory indicates that the dichotomy in experimental rat studies may be caused by the use of insufficiently validated estrogen administration methods resulting in serum hormone concentrations far from those intended, and that physiological estrogen concentrations are neuroprotective while supraphysiological concentrations augment the damage from cerebral ischemia. This evidence offers a new perspective on the mechanisms of estrogens' actions in cerebral ischemia, and also has a direct bearing on the hormone replacement therapy debate. Estrogens affect their target organs by several different pathways and receptors, and the mechanisms proposed for their effects on stroke probably prevail in different concentration ranges. In the current article, previously suggested neuroprotective and neurodamaging mechanisms are reviewed in a hormone concentration perspective in an effort to provide a mechanistic framework for the dose-dependent paradoxical effects of estrogens in stroke. It is concluded that five protective mechanisms, namely decreased apoptosis, growth factor regulation, vascular modulation, indirect antioxidant properties and decreased inflammation, and the proposed damaging mechanism of increased inflammation, are currently supported by experiments performed in optimal biological settings.