The impact of repeated and prolonged attempts at remembering on false memory rates was assessed in three experiments. Participants saw and imagined pictures and then made repeated recall attempts before taking a source memory test. Although the number of items recalled increased with repeated tests, the net gains were associated with more source errors (i.e., calling imagined items ''seen''). Source errors were particularly marked when earlier memory tasks involved remembering without consideration of the source of the remembered items, and occurred for both semantically related and physically similar items. Source misattributions occurred on both forced and free recall tests, thus were not merely due to guessing or changes in response criteria. Similar results were found when a prolonged recall period was used instead of separate memory tests. Hence, source confusions can increase when people reflect on their memories repeatedly or over prolonged periods of time without carefully considering their source.