In two experiments, effects of incentives on task switching were investigated. Incentives were provided as a monetary bonus. In both experiments, the availability of a bonus varied on a trial-to-trial basis. The main difference between the experiments relates to the association of incentives to individual tasks. In Experiment 1, the association of incentives to individual tasks was fixed. Under these conditions, the effect of incentives was largely due to reward expectancy. Switch costs were reduced to statistical insignificance. This was true even with the task that was not associated with a bonus. In Experiment 2, there was a variable association of incentives to individual tasks. Under these conditions, the reward expectancy effect was bound to conditions with a well-established bonus-task association. In conditions in which the bonus-task association was not established in advance, enhanced performance of the bonus task was accompanied by performance decrements with the task that was not associated with a bonus. Reward expectancy affected mainly the general level of performance. The outcome of this study may also inform recently suggested neurobiological accounts about the temporal dynamics of reward processing.