In task switching studies, pre-cuing of the upcoming task improves performance, indicating preparatory activation of the upcoming task-set, and/or inhibition of the previous task-set. To further investigate cue-based task preparation, the authors presented both valid and invalid task cues in a task switching experiment involving three tasks. Consistent with previous findings, a validity effect in terms of higher reaction times on invalidly compared to validly cued tasks was obtained. However, this validity effect was reduced following invalidly cued trials, suggesting dynamic adjustment in terms of decreased cue-based preparation after being misled. Performance was particularly impaired when the current task was the one that was invalidly cued on the preceding trial. This finding may reflect either particular reluctance to prepare or persisting inhibition of the erroneously prepared task-set from the pre-trial.