Remembering to do something in the future (termed prospective memory) is distinguished from remembering information from the past (retrospective memory). Because prospective memory requires strong self-initiation, Craik (1986) predicted that age decrements should be larger in prospective than retrospective memory tasks. The aim of the present study was to assess Craik's prediction by examining the onset of age decline in two retrospective and three prospective memory tasks in the samples of young (18-30 years), young-old (61-70 years), and old-old (71-80 years) participants recruited from the local community. Results showed that although the magnitude of age effects varied across the laboratory prospective memory tasks, they were smaller than age effects in a simple three-item free recall task. Moreover, while reliable age decrements in both retrospective memory tasks of recognition and free recall were already present in the young-old group, in laboratory tasks of prospective memory they were mostly present in the old-old group only. In addition, older participants were more likely to report a retrospective than prospective memory failure as their most recent memory lapse, while the opposite pattern was present in young participants. Taken together, these findings highlight the theoretical importance of distinguishing effects of ageing on prospective and retrospective memory, and support and extend the results of a recent meta-analysis by Henry, MacLeod, Phillips, and Crawford (2004).