Prospective memory (PM) is defined as remembering to remember or remembering to perform an intended action, which requires to form and later realize intentions that must be delayed for minutes, hours, or days, such as remembering to take medication with a meal, or to turn off the stove after cooking. Prospective memory is important in our everyday life, especially to those old people with declined (i.e., less efficient) cognitive functions. Better prospective memory performance can help old people maintain their independent lives and improve their quality of life. So the present study focused on the age difference in an event-based prospective memory task. Regarding the internal mechanism of age differences in PM, two contradictive perspectives were proposed: the age deficits exist in prospective component VS in retrospective component. Noticing-search model (Einstein McDaniel, 1996) suggested that the PM deficit in old people came from the searching process. On the other hand, Smith and Bayen (2006) applied a formal multinomial processing tree model of PM and indicated that the deficit existed in the process of recognizing target cues, which was resource-demanding. The present study evaluated the two perspectives and further investigated the related cognitive mechanism using a classic paradigm where the prospective memory task was embedded in a semantic categorization task. 60 young (29 females) and 60 old adults (26 females) participated experiment 1 voluntarily. The experiment compared the effect of age on the prospective component and the retrospective component of PM, manipulating semantic relatedness between cue and intention, and the distinctiveness of the cue. Young adults performed better than old adults in both the prospective component and retrospective component, but the effect size of age was larger in prospective component. When the relatedness was high or the cue was distinctive, old adults performed as well as young adults, however, it showed age deficits when the relatedness was low or the cue was indistinguishable.These findings suggest that both data-driven and conceptually driven processes can improve old people's memory for intentions greatly.Another 68 young (37 females) and 68 old adults (46 females) participated in Experiment 2. The experiment examined the effect of prospective component on the age difference in prospective memory by reducing the demand of retrospective component and manipulating the cognitive resources of prospective memory. Main effect of age, the distractive level of divided attention and the distinctiveness of the target cue were all found in the accuracy of prospective memory. Both the response time of ongoing and PM tasks had a significant decline in the older adults compared to the young ones. Significant age deficits were found when the target was indistinguishable. And the high distractive level task made old adults’ performance even worse. To be summarized, the aging of PM is modulated both by data-driven and conceptually driven processes. Compared to retrospective component, the age effect in PM is greater on the prospective component. The cognitive load in the task influences the age effect in PM since the prospective component is resource-demanding. The age effect is more obvious when the PM task needs more strategic control than automatically processed. These results further reveal the cause of the reasons for age differences in prospective memory.