Childhood is a period of pronounced improvements in children's ability to remember connections between details of an event (i.e. binding ability). However, the mechanisms supporting these changes remain unclear. Prior evidence is mixed, with some proposing that improvements in the ability to identify previous connections (i.e. increases in hits) account for memory changes, whereas other evidence suggests changes are additionally supported by the ability to identify inaccurate connections (i.e. decreases in false alarms). To disentangle the role of each process, we investigated changes in hits and false alarms within the same paradigm. The present study of 200 4-to-8-year-old children (100 female) used a cohort sequential design to assess longitudinal change in binding ability. Developmental trajectories of d', hit, and false alarm rates were examined using latent growth analysis. Findings demonstrated non-linear improvements in children's binding ability from age 4-to-8-years. Improvements were differentially supported by hits and false alarms. Hit rates improved non-linearly from 4-to-8-years, with greater growth from 4-to-6-years. False alarm rates did not significantly change from 4-to-6-years, but significantly decreased from 6-to-8-years. Overall, findings show improvements in binding ability are predominantly supported by increased hit rates between 4-to-6-years and by both increasing hit rates and decreasing false alarms rates between 6-to-8-years. Together, these results suggest that binding development is non-linear and that mechanisms underlying improvements differ across childhood.