Judgments of Learning (JOL) that are made after a delay, instead of immediately after study, are more accurate in terms of predicting later recall (the delayed JOL effect). The Self-Fulfilling Prophecy (SFP) theory explains the delayed JOL effect as the result of a testing effect. In the current study we tested the prediction that performing delayed JOLs leads to a memory improvement. During learning, 79 participants studied Swahili-Swedish word pairs, immediately followed by a cued recall test, and then made either one single or three repeated, spaced JOLs. A final cued recall test was given after either 5 minutes or 1 week. Making repeated JOLs did not increase memory performance compared to the single JOL condition, hence lending no support to the SFP theory. However, making repeated JOLs did improve their relative accuracy, which suggests that the delayed JOL effect mainly concerns memory monitoring and not performance.