How do we assess what we remember? Previous work on metacognition suggests that confidence judgments are more accurate when given after than before a response to a perceptual task. Here we present two experiments that investigate the influence of decision and response on metacognitive accuracy in a memory task so as to establish what kind of information people use to assess their memory content. Participants were asked to remember lists of words and then to decide which of two target words had previously been presented. In both experiments, participants rated their confidence either after or before the response. However, the experiments differed in the amount of information provided for confidence rating. In Experiment 1, before confidence rating, participants were either presented with both target words and asked to decide between them, or they were only presented with a cue (first letter of the subsequent target words). In Experiment 2, participants were always presented with a target word before confidence rating. The results of both experiments showed that although task accuracy correlated with confidence ratings in both conditions, this relationship was weaker when confidence was assessed before response to a memory task. We argue that metacognitive judgments are influenced by processing information that is not available at the time of primary response. We discuss the implications for theories of confidence and metacognition.