There are limited therapeutic options for treating acute ischaemic stroke. Two approaches have received considerable attention, thrombolysis and neuroprotection. Thrombolysis has provided the only clinically approved drug, recombinant tissue plasminogen activator (rtPA). However patients must undergo a scan before its administration, it must be given within 3 h of symptom onset, and it increases the risk of haemorrhagic complications. Consequently no more than 5% of patients receive rtPA. New thrombolytics and anti-aggregation compounds are in development, including desmoteplase, a genetically engineered version of a thrombolytic protein in Vampire Bat saliva. Preliminary evidence suggested that desmoteplase-treated patients had a higher rate of early reperfusion, better clinical outcome and no increase in haemorrhage rate compared to the placebo group. Another approach to increase cerebral blood flow is the defibrinogenating compound ancrod. Success with these compounds in current Phase III trials has to encompass not only improved functional outcomes compared to placebo but also a longer therapeutic window than rtPA. These compounds would also have to possess a satisfactory adverse event profile, particularly a low incidence of hemorrhagic transformation, a problem that has limited clinical acceptance of rtPA. Neuroprotection is an entirely different approach. The role of a neuroprotectant is to interfere with the biochemical chain of events that results in neurodegeneration, thereby limiting tissue damage. The basis of neuroprotection is that the area of markedly reduced blood flow (the core) is surrounded by the penumbra which, while compromised, can be protected either by reflow or administration of a neuroprotectant. Without intervention cells in the penumbra die. Over the last 20 years huge efforts have been expended in developing neuroprotectants. Around 1000 compounds have been developed preclinically with many showing efficacy in preclinical stroke models. Over 100 reached clinical evaluation but none has clearly demonstrated efficacy. Because of these many failures a group of experts formulated a set of guidelines to be used for preclinical drug development before a compound progressed to the clinic (Stroke Therapy Academic Industry Roundtable, 1999). At the time the guidelines were published the novel free radical trapping neuroprotectant NXY-059 was being developed and its preclinical programme was focussed on following the guidelines which included full dose-response information, validation in permanent focal ischemia models, short and long term functional outcome measures and confirmation of results in non-company academic centres. NXY-059 fulfilled all criteria (Green & Ashwood, 2005) and was therefore examined in acute stroke patients. A first Phase III trial (1700 patients) was positive on the primary end point (Lees et al. 2006) but a second trial (3200 patients) failed to confirm efficacy. This result both emphasises the importance of conducting more than one large trial in order to prove clinical efficacy and, crucially, questions the validity of the STAIR criteria. There remain 2 compounds in late clinical development, DP-b99 and citicoline, but the outcome of their clinical efficacy remains uncertain, given the apparent lack of predictive value of current preclinical evaluation techniques. Why has success for the neuroprotection approach been elusive? Possibly because we are not modelling the disease closely enough. Many stroke patients are old, with co-morbid conditions such as hypertension and diabetes and these are not being modelled when young healthy animals are used. Such co-morbid conditions may alter cerebrovascular functions such as the structure of the blood brain barrier or the neuroimmune system. The future of neuroprotection research may require better animal models. Alternatively, the failure of NXY-059, a compound developed with close regard to the best established guidelines of preclinical and clinical methodology, may result in pharmaceutical companies switching strategies. Few companies will feel the huge costs of a neuroprotectant clinical trial programme are justified in the light of so many failures. Consequently we are likely to see increasing research and development on approaches which involve neuro-restoration and repair in the region of the cerebral infarct. There is already much information on the mechanisms associated with cell death and cell survival in the CNS. Manipulation of such mechanisms may prove valuable following a stroke. What remains uncertain is how one might “switch on” or “switch off” such mechanisms in the damaged brain without disrupting homeostasis in the many other regions of the organism where such intervention would almost certainly prove problematic.