Performing consecutive self-control tasks typically leads to deterioration in self-control performance. This effect can be explained within the strength model of self-control or within a cognitive control perspective. Both theoretical frameworks differ in their predictions with regard to the impact of affect and task characteristics on self-control deterioration within a two-task paradigm. Whereas the strength model predicts decrements in self-control performance whenever both tasks require a limited resource, under a cognitive control perspective, decrements should only occur when people switch to a different response conflict in the second task. Moreover, only the cognitive control model predicts an interaction between task switching and positive affect. In the present research, we investigated this interaction within a two-task paradigm and found evidence that favored a cognitive control interpretation of the results. Positive affect only benefitted consecutive self-control performance if response conflicts in the two tasks were different (resisting sweets followed by a Stroop task). If they were the same (two consecutive Stroop tasks), positive affect impaired self-control performance. These effects were partially replicated in the second study that also examined negative affect, which did not affect self-control performance. We conclude that drawing on cognitive control models could add substantially to research on self-control.