The explicit task-cuing procedure is commonly used to study executive control processes involved in set switching, but performance in this task-switching procedure may be accomplished without switching tasks. Subjects may perform both tasks by using a compound-stimulus strategy, in which subjects encode the cue, encode the target, and use the combination as a compound retrieval cue to choose the appropriate response. We manipulated the number of targets (8,16,32, or 640) that subjects experienced in a four-cue/two-task procedure to separate episodic and semantic memory retrieval components of the compound-stimulus strategy. Cue repetitions were faster than task repetitions, and task repetitions were only slightly faster than task alternations, suggesting that cue repetition effects account for the bulk of the difference between repetitions and alternations. We found the same effects with all target set sizes. The results are consistent with use of a semantic compound-stimulus strategy.