Patients with mild Alzheimer's disease (AD) and age-matched controls were compared on a series of tasks designed to measure errors of mis-attribution, the act of attributing a memory or idea to an incorrect source. Mis-attribution was indexed through the illusory truth effect, the tendency for participants to judge previously encountered information to be true. Cognitive theories have suggested that the illusory truth effect reflects the mis-attribution of experimentally produced familiarity (a nonspecific sense that an item has been previously encountered) to the veracity of previously encountered information. Consistent with earlier suggestions that AD impairs both familiarity and recollection (specific memory for contextual details of the study episode), AD patients demonstrated significantly fewer mis-attribution errors under conditions in which the illusory truth effect is thought to rely on relative familiarity (uncued condition), but more mis-attribution errors under conditions thought to rely on relative amounts of contextual recollection (cued condition). These results help further specify the precise nature of memory impairments in AD.