Two experiments investigated the effects of normal aging and dementia on laboratory-based prospective memory (PM) tasks. Participants viewed a film for a later recognition memory task. In Experiment 1, they were also required either to say "animal" when an animal appeared in the film (event-based PM task) or to stop a clock every 3 min (time-based PM task). In both tasks, young participants were more successful than older participants, who were, in turn, more successful than patients with Alzheimer's disease (AD). For successful remembering in the time-based task, older participants and AD patients checked the clock more often than did young participants. In Experiment 2, participants were asked to reset a clock either when an animal appeared in the film (unrelated cue-action) or when a clock appeared in the film (related cue-action). Responses were faster in the related condition than in the unrelated condition. Again, there were differences in PM performance between young and older participants, and between older participants and AD patients. The observed deficits were not due to the forgetting of the PM task instructions in either experiment. Retrospective memory (RM) tasks (digit span, sentence span, free recall, and recognition) were more impaired by AD than were the PM tasks. Factor analysis revealed separate factors corresponding to RM and PM.