Horton and Gerrig (2005a) outlined a memory‐based processing model of conversational common ground that provided a description of how speakers could both strategically and automatically gain access to information about others through domain‐general memory processes acting over ordinary memory traces. In this article, we revisit this account, reviewing empirical findings that address aspects of this memory‐based model. In doing so, we also take the opportunity to clarify what we believe this approach implies about the cognitive psychology of common ground, and just as important, what it does not imply. We also highlight related areas of research demonstrating how general cognitive processes can constrain access to relevant knowledge in ways that shape both language production and comprehension.