A central issue in the research of directed forgetting is whether the differential memory performance for to-be-remembered (TBR) and to-be-forgotten (TBF) items is solely due to differential encoding or whether retrieval inhibition of TBF items plays an additional role. In this study, recognition-related event-related brain potentials (ERPs) were used to examine this issue. The spatio-temporal distributions of the old/new ERP effects obtained in Experiment 1 that employed a directed forgetting paradigm were compared with those recorded in Experiment 2 in which the level of processing was manipulated. In Experiment 1, participants were instructed to remember or to forget words by means of a cue presented after each word. ERPs recorded in the recognition test revealed early phasic frontal and parietal old/new effects for TBR items, whereas TBF items elicited only a frontal old/new effect. Moreover, a late right-frontal positive slow wave was more pronounced for TBF items, suggesting that those items were associated with a larger amount of post-retrieval processing. In Experiment 2, the same cueing method and the same stimulus materials were used, and memory encoding was manipulated by cueing participants to process the words either deeply or shallowly. Both deeply and shallowly encoded items elicited phasic frontal and parietal old/new effects followed by a late right-frontal positive slow wave. However, in contrast to TBR and TBF items, these effects differed only quantitatively. The results suggest that differential encoding alone cannot account for the effects of directed forgetting. They are more consistent with the view that items followed by an instruction to forget become inhibited and less accessible, and, therefore, more difficult to retrieve.