Prospective memory (PM) involves remembering intended actions in the future, such as posting a letter when seeing a post box (event-based PM) or making a phone call at 2:00 pm (time-based PM). Studies on aging and PM have often reported negative age effects in the laboratory, but positive age effects in naturalistic tasks outside the laboratory (the so-called age–PM-paradox). The present study re-examined this pattern of the paradox by studying, for the first time, age differences in time- and event-based PM in lab-based, experimenter-generated naturalistic and self-assigned real-life PM tasks within the same sample of young and older adults. Results showed that differential age effects in and outside the laboratory were qualified by the type of PM cue. While age-related deficits were obtained for laboratory event-based tasks, no age effect was obtained for naturalistic event-based PM. Age benefits in the field were only observed for naturalistic time-based tasks, but not for participants’ own self-assigned time-based tasks. These findings indicate that the age benefits for naturalistic PM tasks may have been overestimated due to the dominant use of experimenter-generated naturalistic time-based PM tasks in previous studies. Therefore, the precise pattern of the age–PM-paradox may need redefining as mostly consisting of negative age effects in lab-based PM tasks and mostly the absence of negative age effects (rather than age benefits) in naturalistic and self-assigned tasks outside the laboratory.