Forgetting irrelevant/problematic actions in a domain description can be useful in solving reasoning problems, such as query answering, planning, conflict resolution, prediction, postdiction, etc.. Motivated by such applications, we study what forgetting is, how forgetting can be done, and for which applications forgetting can be useful and how, in the context of reasoning about actions. We study these questions in the action language \cal C (a formalism based on causal explanations), and relate it to forgetting in classical logic and logic programming.