Stroke models in animals, particularly in rodents have limited translational value for several resons. E.g. stroke in humans has a more complex physiopathology and very often different comorbidities that play a role. Another difference is the time window for therapeutic intervention that in man is very short at variance with animal models. Also small strokes in sensible areas can determine consequences more severe than larger strokes in less sensible areas and this is something difficult to reproduce in animals. Anyway as there are no real alternatives different models in rodents using different protocols in terms of choice of artery ligation and timing and duration of intervention may help in having a better knowledge on the therapeutic potential. A partial alternative is to use bigger animals, e.g. rabbits or dogs but the predictive power ramains low and the amount of drug substance needed increases significantly and this can create problems in early stage of development. I guess you may find very useful the article at the link below (free full text available).