Assessment of amnesia in Dissociative Identity Disorder (DID) typically relies on self-report, the veracity of which cannot often be independently verified. Memory in DID was therefore assessed using an objective method that involved event-related potentials (ERPs) as well as indirect behavioral measures of memory, and that provided statistically supported assessments for each participant. Four participants who met DSM-IV criteria for DID participated in an ERP memory assessment task, in which words learned by one identity (identity A) were then presented to a second identity (identity B). All four participants - tested as identity B - produced ERP and behavioral evidence consistent with recognition of the material learned by identity A. While it would be premature to generalize all cases of DID, the results suggest that there may be reasons to question the veracity of reports by individuals who meet diagnostic criteria for DID on the basis of a structured clinical interview.