Source memory is typically better when a piece of information is schematically unexpected for its source (e.g., microwave in a bathroom) than when it is expected (e.g., microwave in a kitchen, e.g., Schaper et al., 2019). This inconsistency effect is explained by enhanced attention to unexpected source-item pairs which improves encoding (e.g., Küppers & Bayen, 2014). In metamemory, however, people show an expectancy illusion: They incorrectly predict better source memory for expected source–item pairs and neglect these during study (Schaper & Bayen, 2021). We aimed to improve memory for expected sources by remedying the expectancy illusion and improving metamemory control. Two participant groups (n = 72 each) completed a source-monitoring experiment with expected and unexpected source-item pairs. Participants in the remedy group experienced the inconsistency effect during a first study-test cycle. This remedied their expectancy illusion. For a second study phase (self-paced, new materials), they were instructed to focus their attention on the expected source-item pairs. Finally, they completed a second source-monitoring test. The control group completed only one self-paced study phase and one source-monitoring test without further study instruction. Control participants studied unexpected pairs longer than expected pairs and showed the typical inconsistency effect on source memory. By contrast, in the second study-test cycle of the remedy group, study time was equal for expected and unexpected pairs, as was source memory. Metamemory control, when effectively executed, can improve memory for expected sources.