In the task-switching paradigm, mixing costs indicate the performance costs to mix two different tasks, while switch costs indicate the performance costs to switch between two sequentially presented tasks. Applying tasks with bivalent stimuli and responses, many studies demonstrated substantial mixing and switch costs and a reduction of these costs as a result of practice. The present study investigates whether extensive practice of a task-switching situation including tasks with univalent stimuli eliminates these costs. Participants practiced switching between a visual and an auditory task. These tasks were chosen because they had shown eliminated performance costs in a comparable dual-task practice study (Schumacher et al. Psychol Sci 12:101-108, 2001). Participants either performed the tasks with univalent responses (i.e., visual-manual and auditory-verbal stimulus-response mappings) or bivalent responses (i.e., visual-manual and auditory-manual stimulus-response mappings). Both valence conditions revealed substantial mixing and switch costs at the beginning of practice, yet, mixing costs were largely eliminated after eight practice sessions while switch costs were still existent.