Three experiments that document a powerful false memory effect in recall and recognition are reported. Subjects studied Lists of items related to critical nonpresented themes (e.g., sleep, needle) and then completed recognition and/or recall tests, as well as various metamemory tasks (e.g., remember/know, source monitoring). Results showed that the critical nonpresented items were recalled and recognized nearly as often as studied items. False recognition of critical nonpresented items did not decrease over a 24-h retention interval, whereas the recognition rate for studied items did decrease significantly. The false memory effect also increased across successive recall tests. Subjects' responses on the metamemory tasks indicated that they experienced the critical nonpresented items as being very similar to the presented items that were recalled and/or recognized. Subjects were even willing to indicate which of two persons speaking the study items had spoken the critical nonpresented items that were recalled. Results are discussed in terms of fuzzy trace theory and it is argued that these false memory effects represent memory illusions that may yield important insights into normal human memory processes. (C) 1996 Academic Press, Inc.