Substantial efforts are being made to develop drugs which will protect the brain from the neurodegeneration that follows an acute ischaemic stroke. However, while there are already a significant number of animal models of stroke, there is currently no information as to whether activity of a compound in any of them will predict clinical efficacy. In this article, Jackie Hunter, Richard Green and Alan Cross review the major models of acute cerebral ischaemia and propose rational protocols for examining novel neuroprotective agents.