In this study we investigated the effect of predictability on dual-task performance in two experiments. In the first experiment 33 participants separately practiced a continuous tracking task and an auditory reaction time task. Both tasks had a repeating element that made them predictable; in the tracking task this was a repeating segment, and in the auditory task this was an auditory sequence. In addition, one group obtained explicit knowledge about the repeating sequence in the tracking task while the other group trained implicitly. After training, single-and dual-task performance was tested at a post test and retention test. Results showed that predictability only improved performance in the predictable tasks themselves and dual-task costs disappeared for the tracking task. To see whether the task-specific effect of predictability was the results of task prioritization, or because task representations did not have much chance to interact with each other, we conducted a second experiment. Using the same tasks as in Experiment 1, 39 participants now trained both tasks simultaneously. Results largely mirrored those of the first experiment, demonstrating that freed-up resources due to predictability in one task could not be re-invested to improve in the other task. We conclude that predictability has a positive but task-specific effect on dual-task performance.