In many task-switch studies, task sequence and response sequence interact: Response repetitions produce benefits when the task repeats but produce costs when the task switches. Four different theoretical frameworks have been proposed to explain these effects: a reconfiguration-based account, association-learning models, an episodic-retrieval account, and a priming and inhibition account. The main goal in the present study was to test the unique prediction of the priming and inhibition account that stimulus categories remain active from one trial to the next, thus counteracting the negative effects of response inhibition in task-repetition trials. As testing this prediction required a somewhat untypical task-switch design, a second aim in the present study consisted in evaluating the generality of the alternative models. In the present experiments the task-switch paradigm was modified to include trials in which pure stimulus-category repetitions could occur. Across 3 experiments, benefits were observed for stimulus-category repetitions in task-switch trials, a prediction that conforms only to the priming and inhibition account and the reconfiguration account. However, the benefits in task-repetition trials were consistently smaller than the benefits in task-switch trials. This effect is in line only with the predictions from the priming and inhibition account. Thus, the current results support the notion of stimulus-category priming and response inhibition as the 2 mechanisms causing the opposite response-repetition effects in task-repetition and task-switch trials.