Two experiments examined the role of predictability within the elements of a task-irrelevant auditory sequence on the disruption produced to visual-verbal serial recall. Experiment 1 showed that participants did not benefit from having a long-term representation of the irrelevant sequence: A highly predictable, canonical sequence (“1 2 3 4 5 6 7 8 9”) produced as much disruption as a repeated random sequence (which was the same on each trial) and an unpredictable, random sequence (which differed on each trial), as compared with quiet. In line with this finding, there was also no difference between a predictable canonical and an unpredictable random sequence in Experiment 2. However, a deviant within the predictable, canonical sequence (“1 2 3 4 5 5 7 8 9”) produced greater disruption than a deviant within an unpredictable, random sequence (“4 8 2 9 5 5 7 3 1”). This effect was confined to early trials within the block. The results showed that long-term knowledge about the order of the individual elements in the sequence did not help attenuate the effect of auditory distraction on serial recall. Nevertheless, attentional capture was amplified when a deviant violated a well-known, canonical sequence, providing evidence that the neural model represents postcategorical sequential information.