Previous studies have indicated that older adults have a special deficit in the encoding and retrieval of associations. The current study assessed this deficit using ecologically valid name-face pairs. In two experiments, younger and older participants learned a series of name-face pairs under intentional and incidental learning instructions, respectively, and were then tested for their recognition of the faces, the names, and the associations between the names and faces. Under incidental encoding conditions older adults' performance was uniformly lower than younger adults in all three tests, indicating age-related impairments in episodic memory representations. An age-related deficit specific to associations was found under intentional but not under incidental learning conditions, highlighting the importance of strategic associative processes and their decline in older adults. Separate analyses of hits and false alarms indicate that older adults' associative deficit originated from high false alarm rates in the associative test. Older adults' high false alarm rates potentially reflect their reduced ability to recollect the study-phase name-face pairs in the presence of intact familiarity with individual names and faces.