The current study explores the role of three components of working memory in age differences in an executive task, the Tower of London (TOL). The TOL task is sensitive to frontal lobe damage, and is widely used to measure planning ability. Dual tasks were used to test the involvement of the phonological loop (articulatory suppression), visuospatial buffer (pattern tapping), and central executive (random generation) in age effects on the TOL. Older adults showed greater reliance than young on domain-specific verbal and spatial memory components in performing the TOL. In terms of executive function, qualitatively different interference patterns were seen in young and old participants. However, the validity of using random generation tasks to assess executive function in older populations can be questioned. For older participants, performing the TOL loads all components of working memory, whereas for the younger participants the TOL more specifically loads executive functioning.