Performing action phrases (subject-performed tasks, SPTs) leads to better memory than verbal learning instructions (verbal tasks, VTs). In Experiments 1-3, the list-method directed forgetting design produced equivalent directed forgetting impairment for VTs and SPTs; however, directed forgetting enhancement emerged only for VTs, but not SPTs. Serial position analyses revealed that both item types suffered equivalent forgetting across serial positions, but enhancement was evident mostly in the first half of List 2. Experiment 4 used the item-method of directed forgetting and obtained greater directed forgetting for VTs than SPTs. A "remember-all" baseline group allowed estimating the impairment for to-be-forgotten (TBF) items and enhancement for to-be-remembered (TBR) items. Serial position analyses showed greater impairment for TBF items from the beginning of the list than elsewhere in the list. Directed forgetting enhancement for TBR items occurred throughout the list for VTs, but only in the primacy region for SPTs. Overall, dissociations across the list-method and item-method studies with SPTs suggest that the two methods have different underlying mechanisms. Furthermore, dissociations obtained with SPTs "within" list-method studies provide support for the dual-factor directed forgetting account and challenge the single-factor accounts.