In recent years, there have been calls for researchers to report and interpret confidence intervals (CIs) rather than relying solely on p-values. Such reforms, however, may be hindered by a general lack of understanding of CIs and how to interpret them. In this study, we assessed conceptual knowledge of CIs in undergraduate and graduate psychology students. CIs were difficult and prone to misconceptions for both groups. Connecting CIs to estimation and sample mean concepts was associated with greater conceptual knowledge of CIs. Connecting CIs to null hypothesis significance testing, however, was not associated with conceptual knowledge of CIs. It may therefore be beneficial to focus on estimation and sample mean concepts in instruction about CIs. [This article was published in "Statistics Education Research Journal" (EJ1217507).]