One of the most striking features emerging from studies of adult age differences in human memory is that loss of ability is very large under some conditions, but small or non-existent under others. This pattern of findings suggests a number of important questions: First, what are the reasons underlying these apparent discrepancies? Second, do the discrepencies provide a clue to the reasons for memory failures associated with aging? Third, does the pattern of results provide clues to the nature of memory itself? The purposes of the present chapter are to review some recent empirical work fitting the pattern described above, and to give preliminary answers to the questions suggested by the data.