Firing the manager is a drastic measure employed by firms to deal with poor performance. However, data on within-firm dynamics are scarce, and the firing of managers is rarely recorded in the firm level data currently available. This makes the value of firing a manager difficult to assess. Data on sports offer a unique opportunity to study this phenomenon. The firing of a coach is usually well-publicized, and therefore an observable event. Using data on soccer, the author evaluates the effect of the firing of a coach on team performance. As teams do not face the same opponents before and after a coach is fired, issues of sample selectivity are addressed. Keywords: professional soccer, rating models, Poisson model