Prior prospective memory (PM) research shows paradoxical findings-young adults outperform older adults in laboratory settings, but the reverse is found in naturalistic settings. Moreover, young-old outperform old-old adults in laboratory settings, but show no age differences in naturalistic settings. Here we highlight how time-based task characteristics have differed systematically between studies conducted in laboratory (time-interval cues) and naturalistic settings (time-of-day cues) and argue that this apparent paradox is a function of comparing disparate task types. In three experiments, we tested this hypothesis using analogous paradigms across settings, with event-based, time-of-day, and time-interval cued PM tasks. Experiment 1 compared young (n = 40) and older (n = 53) adults on a laboratory paradigm that measured PM tasks embedded in a virtual, daily life narrative; and on a conceptually parallel paradigm using a customized smartphone application (MEMO) in actual daily life. Results revealed that on the MEMO, older adults outperformed young adults on the time-of-day tasks but did not differ on the time-interval or event-based task. In contrast, older adults performed worse than young adults in the laboratory. Experiment 2 compared PM performance in young-old (n = 64) and old-old (n = 40) adults using the same paradigms. Young-old outperformed old-old adults in the laboratory; however, group differences were not evident in daily life. Experiment 3 compared young (n = 42) and older (n = 41) adults, and largely replicated the findings of Experiment 1 using a more demanding version of MEMO. These findings provide novel and important insights into the limiting conditions of the age-PM paradox and the need for a finer theoretical delineation of time-based tasks.