People believe most background sounds to disrupt their cognitive performance. An exception is music they like which is believed to improve cognitive performance. To examine the objective effects of music on cognitive performance, the serial-recall paradigm was used. Mozart's sonata K. 448 - the music piece used in classical studies on the Mozart effect - caused distraction. However, with ongoing exposure to the music participants were able to adapt to some degree to the distracting effect of the music under conditions that increased the predictability of the sounds, suggesting that the development of a predictive model of the unfolding auditory input helped to diminish distraction. Retrospective metacognitive judgements indicated that the more participants liked the music the more likely they were to judge that it had helped their performance. However, serial recall was disrupted by the music irrespective of whether the music was liked or not.