This paper aims to review the literature on event studies. It highlights the importance of the methodology used for estimating long-run abnormal returns, as event studies are sensitive to the returns generating process in this time horizon (Savickas, 2003, and Aktas et al., 2007). We find that the impact of events has to be controlled for in the long run without regard to the estimation of the window for abnormal returns, in which the best estimation methodology, in terms of robustness, turns out to be the two-state market model.