The discrepancy-reduction model of learning holds that accuracy of metacognitive monitoring affects learning through its influence on self-regulated study. Thiede, Anderson, & Therriault (2003) have underscored the importance of efforts to improve monitoring accuracy by showing that more accurate monitoring can result in increased learning. In their study, a group who summarized texts after a delay had better monitoring accuracy, regulated their study better, and learned texts better than groups who either didn't summarize texts or who did so immediately after reading. This effect was partially replicated in the current investigation. In addition, this study shows that working memory capacity affects learning, but has no effect on monitoring accuracy or self-regulated study. Those with large working memory spans learned texts better than those with smaller spans, but were no better at monitoring their learning or at regulating their study.