Directed forgetting occurs when people show poorer recall for information associated with a forget instruction than information associated with a remember instruction. In a recent extension directed forgetting, called selective directed forgetting, the instruction is to forget only a subset of what is presented. Research on selective directed forgetting has yielded mixed results. The aim of this dissertation is to clarify these results by examining the effectiveness of a selective forget instruction when the to-be-forgotten information is contained in an event model. An event model is a mental representation that contains elements of an episode such as when and where it took place. In the experiments, participants read a series of sentences and were instructed to forget some of them. The to-be-forgotten sentences could or could not be integrated into event models. The results suggest that selective forgetting is possible but only when the to-be-forgotten information does not have a common structure (i.e., is not related by theme or event). This suggests that people do have some amount of control over the contents of their memories.