Sixty-one participants were asked (a) to recall a memory for a period lasting 15 minutes and (b), at the end of this period, to estimate retrospectively the duration of this period. They were assigned to one of four groups: the memory was either joyful or sad, and was recent (within the past two years) or old (when the participant was 7 to 10 years old). The most critical finding is the demonstration that the age of the recalled memory has an impact on the verbal estimation. More specifically, duration is underestimated in the old but not in the recent memory condition. Moreover, in this study, recalling a memory, old or recent, is shown to be an efficient way to generate a joyful or sad emotion. Finally, the results also indicate that there is a significant correlation between the uncertainty related to the duration estimated retrospectively and the score on the present-hedonistic scale of the Zimbardo Time Perspective Inventory.