Problems with cognitive control in both autism and dyslexia have already been reported in different studies. The present study specifically examined task-switching performance in children with autism and dyslexia. For this purpose, a multiple-trial paradigm was used with cues for colour- and shape-matching tasks presented before a run of trials. The cue could imply a task switch (when the cue changed the task) or a task repetition (when the cue did not change the task). Both reaction times and error rates were measured for switching, restarting, and general task performance. Participants were children with autism (24) and with dyslexia (25) and healthy controls (27) with normal IQ and ages from 12 to 18 years. The main finding was that while similar switching performance was observed between children with autism and the healthy controls, children with dyslexia showed a significant switch-specific delay relative to both healthy controls and children with autism. Furthermore, no deficit in restarting performance was observed for any of the two patient groups. Finally, additional evidence is provided for a more general deficit in information processing in dyslexia. Our data suggest that children with autism are able to switch between tasks in a similar way as do normally developing children as long as the tasks are unambiguously specified. Furthermore, the data imply switch-specific deficits in dyslexia additionally to the deficits in general information processing already reported in the literature. The implications of our data are further discussed in relation to the interpretation of the Wisconsin Card Sorting Test.