Differences in the impact of irrelevant sound on recall performance in children (aged 7–9 years old; 'N' = 89) compared to adults (aged 18–22 years old; 'N' = 89) were examined. Tasks that required serial rehearsal (serial and probed-order recall tasks) were contrasted with one that did not (the missing-item task) in the presence of irrelevant sound that was either steady-state (a repeated speech token), changing-state (two alternating speech tokens) and, for the first time with a child sample, could also contain a deviant token (a male-voice token embedded in a sequence otherwise spoken in a female voice). Participants either completed tasks in which the to-be-remembered list-length was adjusted to individual digit span or was fixed at one item greater than the average span we observed for the age-group. The disruptive effects of irrelevant sound did not vary across the two methods of determining list-length. We found that tasks encouraging serial rehearsal were especially affected by changing-state sequences for both age-groups (i.e., the changing-state effect) and there were no group differences in relation to this effect. In contrast, disruption by a deviant sound—generally assumed to be the result of attentional diversion—was evident among children in all three tasks while adults were less susceptible to this effect. This pattern of results suggests that developmental differences in distraction are due to differences in attentional control rather than serial rehearsal efficiency.