Evaluative Conditioning (EC), the change in liking towards a neutral stimulus due to its pairing with a positive/negative unconditioned stimulus, is a central effect in attitude formation. Current research emphasizes the role of explicit memory in EC. However, human memory is no passive information-storage device, but people actively monitor and control their own memory processes. In the present research, we examined whether people can monitor their memory processes in attitude formation via EC and let participants predict whether they will remember the stimulus pairings in the future (judgments of learning, JOLs). In six preregistered experiments, higher JOLs were associated with stronger EC effects. Surprisingly, objective memory explained this effect only partially, although people’s JOLs were moderately accurate in predicting memory performance. Following a Brunswikian perspective, we identified several variables contributing to the correlation between JOLs and the EC effect, such as the extremity of the unconditioned stimuli, the fit between conditioned and unconditioned stimuli, and the feeling of processing ease, which all correlated with higher JOLs and stronger EC effects. Further experiments showed the robustness of these effects across different boundary conditions, such as whether judgments and memory tests target the valence or the identity of the stimuli. Our results attest to the role of metamemory in attitude formation via EC, whereby expecting that one will remember a stimulus predicts the size of the EC effect over and above actual memory. By integrating two previously unrelated research areas, our studies provide important theoretical insights into both attitude formation and metamemory.