Many researchers employed secondary tasks, which have to be performed in parallel to a primary task requirement, to examine how successful task selection is accomplished in the context of task switching. The influence of such a secondary task on the observed switch cost (i.e., cost arising when the currently relevant task differs from the task performed most recently) is seen as indicative for involved cognitive processes such as self-instruction. Most secondary tasks so far chosen required a repeated utterance of a word or syllable or a rhythmic movement of the foot. In the present study, we manipulated the dynamics of the to-be performed secondary task (i.e., either being a repeated utterance or movement or a static involvement of muscle activity, like holding a spattle in the mouth). Additionally, we tested for modality-specific influences by asking participants to perform a dynamic or non-dynamic (i.e., static) secondary task with two effector systems, namely oral or manual. Overall, our secondary tasks lead to a reduced (rather than increased) switch cost compared to a control condition without any secondary task. This reduction in switch cost was dependent of secondary-task dynamics but independent of effectors involved, showing larger switch-cost reductions for dynamic secondary tasks. To explain this finding, we suggest that performing secondary tasks interferes with the formation of episodic memory traces that lead to retrieval benefits in case of a task repetition, so that reduced task-switch cost actually represent reduced repetition benefits.