Young children frequently exhibit persistent overconfidence when evaluating their own physical and cognitive abilities. Although task persistence due to overconfidence may be beneficial in some instances, it may also have adverse effects on academic performance. For example, children may reduce cognitive effort because they mistakenly believe that they understand novel tasks. The present work examined two contexts under which young children make more accurate predictions regarding their own ability to recall previously presented items: peer modeling and the number of items to be recalled. Two experiments using the picture recall paradigm demonstrated that both peer modeling and item magnitude may positively impact prediction accuracy and decrease overconfidence in young children. Despite the impact of these interventions on children's recall predictions as a whole, most children were still overconfident in their judgments of learning (JOL). Thus, lending support to our hypothesis that targeted interventions may reduce, but not eliminate, children's overconfidence in their own cognitive abilities.