Previous research has suggested that people’s attachment styles influence memory processes. Most of this work has focused on the encoding and retrieval of information about events that actually took place. The purpose of the present research was to determine (a) whether attachment styles also predict memories for events that never occurred (false memories);(b) whether experimentally induced attachment anxiety leads to the generation of false memories for interpersonal experiences; and (c) whether these errors arise during encoding, maintenance, or retrieval processes. Our results indicated that attachment anxiety is associated with people’s propensities to experience false alarms on recognition tasks for relational stimuli. Moreover, experimentally altering participants’ state levels of attachment anxiety led to more numerous false alarms, as compared with an unprimed control group. These findings are consistent with the idea that attachment-related anxiety might selectively bias and desensitize the encoding of interpersonal events, ultimately leading people to remember events that did not occur. However, experimentally priming anxiety did not lead to more false alarms relative to groups primed with security, raising the possibility that the anxiety-false memory association could be because of making relational issues salient rather than increasing attachment anxiety per se.