It was hypothesized that in a competitive interaction in which an opponent first used threats and fines a) subjects would use the highest level of coercion that could be justified by the available norms, b) subjects would be inhibited by a norm that justified using only low levels of coercion, but c) when the opponent used a magnitude of coercion which exceeded the norm, subjects would also violate the norm and retaliate with an equally high magnitude of coercion, and d) subjects would use the maximum levels of coercion when engaging in a competitive game without norms. These hypotheses were examined in a 3 x 2 factorial experiment in which the amount of coercion allowable by a norm (Low Norm, High Norm, and No Information) and the level of coercion (Low or High) were manipulated. As predicted, aggressive reactions by subjects were inhibited when the provoker conformed to a low norm by using a low level of coercion. On the other hand, subjects retaliated by using a high level of coercion when the level of coercion used by a provoker violated the norm. Subjects who could justify their behavior by the availability of a high punitive norm tended to use all of the coercive power available to them even when the provoker used a low level of Coercion. 0 1996