To better understand the world, humans scan the information contents of our experiences for patterns, corresponding to the extraction of gist, or essential meaning (Brainerd & Reyna, 1990). Research suggests that some forms of gist extraction require several hours, and even sleep for further processing (Ellenbogen et al., 2007; Payne et al., 2009), perhaps becoming subsequently reactivated in the hippocampus (Marshall & Born, 2007). Interestingly, the hippocampus may itself be functionally specialized for gist extraction in its anterior segment (Poppenk et al., 2013). The current study investigated the role of the anterior/posterior hippocampus and sleep stages in predicting patterns of change in gist memory over the course of a week. To assess this link, I identified four types of gist (inferential, statistical, multi-item, and single-item) that were described in recent reviews (Landmann et al., 2014; Stickgold & Walker, 2013). 104 participants were recruited, 67 of whom passed eligibility criteria and completed three behavioural sessions (evening before sleep, 12 hours later in the morning after sleep, and one week after the first session) and an MRI several weeks later as part of a broader battery of tasks. I found evidence that inferential gist in a transitive inference task increased over time, suggesting that new information is being formed. I also found that REM, rather than slow-wave sleep, predicted gist extraction in a number of different tasks. Lastly, hippocampal volumes predicted immediate rather than delayed gist extraction.