Intentional forgetting of previously learned information is an adaptive cognitive capability of humans but its cognitive underpinnings are not yet well understood. It has been argued that it strongly depends on the presentation method whether forgetting instructions alter storage or retrieval stages (Basden, Basden, & Gargano, 1993). In Experiment 1, we compared the processes underlying the directed-forgetting effect in 2 mosts widely used presentation methods, namely the list-method and the item-method, and also differentiated between costs (i.e., poorer memory for to-be-forgotten information) and benefits (i.e., better memory for to-be-remembered information) of directed forgetting within both methods. Using a multinomial modeling approach (Riefer & Rouder, 1992; Rouder & Batchelder, 1998), our results showed that directed-forgetting benefits were due to better storage of to-be-remembered information in both methods. In line with current theorizing, list-method directed-forgetting costs occurred due to reduced retrieval of to-be-forgotten information. Item-method costs, however, occurred not only due to reduced storage, which is the dominant current view, but also due to reduced retrieval. In Experiment 2, we replicated the novel finding that retrieval processes contribute to item-method directed forgetting independent of recall-output order. Implications of these findings for current directed-forgetting theories are discussed.