The fact that people remember more personality trait adjectives in a memory test after making self-reference judgments, relative to other types of judgments (e.g., other-reference and semantics), has been attributed to greater elaborative (i.e., item-specific) and organizational (i.e., relational) processing. The present research asked whether personality self-concept renders more item-specific processing when the self-referenced materials are personality congruent. Previous attempts to address this question were equivocal, as brief recall tests were used and amount recalled was the indicator of item-specific processing. Research shows that item-specific processing produces slower recall than does relational processing, whereas the latter protects against item loss on a second recall test. The present studies therefore tracked recall across two consecutive tests. Results demonstrated an uncrossed double dissociation in self-reference recall speed, wherein self-reported orderliness was associated with slower recall only when the adjectives described orderliness (Study 1; N = 98), and self-reported openness was associated with slower recall only when the adjectives described openness (Study 2; N = 92). In Study 3 (N = 163), participants who viewed their friend as highly orderly recalled orderliness adjectives encoded in an other-reference task more slowly, indicating that tapping the self-concept was not necessary to observe the association. However, the openness-recall speed relationship of Study 2 failed to replicate. In all studies, item loss was not related to personality or recall speed, suggesting relational processing differences were not responsible for the findings. Likewise, demand characteristics were likely not responsible either because personality measures in Study 2 were taken during mass testing. Recall speed was not associated with an indirect personality measure (i.e., Implicit Association Test), suggesting item-specific information was propositional, not associative. Theoretical considerations were centered on whether item-specific information rendered multiple memory traces, which is consistent with the slowing of recall predicted by a random search model of memory and cue-overload theory. Overall, convergent validity was fairly weak and predictive validity was poor, but methodological issues were identified and statistical refinements were suggested. The present studies nevertheless successfully linked individual differences in personality as represented in memory to the item-specific processing that occurred during self- and other-reference encoding.