We examine the strategic behavior of leaders and followers in sequential duopoly experiments in which followers either perfectly observe the leaders' actions or else observe nothing. Our experiments show that consistent with the theory, leaders enjoy a greater …rst-mover advantage when followers observe their actions with higher probability. However, the results also show that (i) leaders do not fully exploit their …rst-mover advantage, (ii) when informed, followers tend to overreact slightly (i.e., choose quantities above their best-response to the leaders' quantities), and (iii) when uninformed, followers try to predict leaders' quantities and react optimally.This suggests that followers view the symmetric Cournot outcome as "fair" and whenever they observe leaders who are trying to exploit their …rst-mover advantage, they "punish" them by overreacting. Such punishments in turn induce leaders to behave more softly than the theory predicts.