Two experiments tested the influences of vivid imagery and person schemata on eyewitness accuracy. Participants watched an event sequence including actors performing stereotype-consistent and inconsistent actions. Additionally, participants either read descriptions of actions (Experiment 1) or vividly imagined actions (Experiment 2). After either 30 minutes or 2 days, recognition memory, source memory, and remember/know judgements were made. After 2 days, false alarms to imagined events increased, relative to the 30-minute test; those false alarms were more often misattributed to stereotype-consistent actors, relative to the same actions in the reading condition. In addition, the accompanying remember judgements were higher for false alarms to imagined events, relative to read events, regardless of stereotype consistency. Overall the results suggest that, over time, vivid imagery reinforces schema activation, increasing stereotype-consistent false memories.