When young children evaluate their confidence, their monitoring is often overoptimistic, that is, inaccurate. The present study investigated a potential underlying mechanism for kindergarteners' and second graders' overconfidence within a paired associates learning paradigm. We implemented a pre-monitoring phase motivating children to differentially evaluate their confidence for each alternative "before" children could choose an answer in the subsequent recognition phase. For one, we intended to weaken the influence of one single and prepotently selected memory trace. For another, we motivated and enabled children to evaluate all four answer alternatives concerning their certainty before evaluating their final recognition choice by giving a confidence judgment. We compared monitoring discrimination and monitoring bias with a control condition whose task sequence did not include a pre-monitoring judgment. Contrary to our expectations, the pattern of results indicated that being instructed to pre-monitor did "increase" and not decrease overconfidence in young children. The present results will be discussed against the background of memory-metamemory interaction, confirmation bias, and methodological issues.