Employing a false alarm recognition procedure with learning of highly associated word pairs, an experiment was conducted to examine the hypothesis of an age-related deficit in the distinctiveness of encoding. The evolution of the false alarm rate and of the C decision criteria was observed across three age groups, young adults, older adults, and older-older adults. The results show 1) no age differences on C decision criteria, indicating that the increase in FA with age is not related to a subject compensation strategy but is probably due to a failure in memory strength, and 2) that older respondents produced significantly more false alarms to distractors related to target items than the young respondents did but that they did not differ in their false alarm rate for unrelated distractors. This finding is interpreted as supporting the hypothesis of a failure with age to encode target items in a sufficiently elaborate or distinctive fashion. For the older-older respondents the data showed an increase in all false alarms indexes, suggesting that the encoding deficit gets worse in late adulthood.