Over the course of six sessions, 24 young (M = 19.40 years, SD 1.61) and 24 older participants (M = 71.48 years, SD 3.86) performed simple, repetitive tapping tasks at 300 and 600 ms target durations concurrently with two cognitive tasks under non-switch or switch conditions. Despite substantial improvements, over sessions, reliable switch costs remained, which were pronounced in older adults. Young and older adults alike showed increased drift in the tapping tasks under dual-task conditions. Under dual-task non-switch conditions, older adults maintained the same timing accuracy (variability) as in the single-task condition. However, variability increased when concurrent cognitive task-set switching was required, while young adults even improved timing accuracy relative to the single-task condition. Being at odds with extant models of timing, our findings demonstrate that control of simple repetitive movements is far from automatic even at intervals below 1 s. Interference with timing in older adults is not caused by multi-tasking per se, but depends on the cognitive control demands of the concurrent task. We argue that our findings suggest a critical role of cognitive control processes for the maintenance of representations of target durations during interval production. This hypothesis received further support from patterns of local interference in the timing of individual intervals.