Working memory’s limited capacity places significant constraints on people's ability to hold information while processing. However, skilled readers are able to effectively encode important information into long-term memory during comprehension. This chapter describes the long-term working memory theory (LT-WM), originally developed to explain how experts in various domains (including reading) enhance their working memory capacity by relying on rapid, skilled use of long-term memory. We first trace the development of the theory and the reasons it took the form it did in the mid-1990s. We explain that LT-WM was not viewed as a new form of memory, but rather as highly practiced use of long-term memory to rapidly and reliably link information together using meaningful associations, retrieval structures, and preexisting knowledge. Next, we describe how the theory accounted for many central phenomena in discourse comprehension. More recent work has proposed a form of LT-WM for syntactic processing as well, and we discuss current critiques of the original evidence advanced to support LT-WM. Finally, we describe recent studies on neural activity associated with LT-WM development in reasoning skills and language comprehension.