This dissertation examines relationships between pedagogical content knowledge for preschool mathematics, math-related teacher language, and gains in children's math ability scores. Pedagogical content knowledge (PCK) is knowledge that, while focused on content, is particularly suited to the purposes of teaching, and specifically tailored to the thinking of the students it is meant to inform. For example, teachers with strong PCK in a content area know which concepts are most central, are aware of the best analogies for illustrating those concepts, and can connect the new ideas they must introduce to the existing interests of their students. Elementary mathematics research has studied PCK in some detail, and provides a framework for an examination of knowledge for teaching preschool mathematics. This study re-evaluates theoretical relationships between mathematical concepts and procedures in the thinking of preschool children, and posits the theory that teacher math-related language input may be particularly effective in supporting early mathematical development. In an empirical application, results from a teacher interview that assesses pedagogical content knowledge (PCK) in preschool mathematics are used to describe the PCK construct in a sample of Head Start teachers in a large midwestern city. Factor analysis suggests the measure detects two underlying constructs: Elaborative PCK and Evaluative PCK. Elaborative PCK is knowledge that supports teachers' efforts to scaffold and elaborate on children's self-directed activities, while Evaluative PCK is knowledge that allows teachers to evaluate children's understandings relative to teacher-imposed learning goals. Analysis using hierarchical linear modeling finds significant relationships between Elaborative PCK and both teacher math-related language and gains in children's learning. While teacher math-related language provided during free play activities also has a significant relationship to child gains (replicating findings of others), the same type of language provided during teacher-structured "circle time" shows no significant association. Implications of this conclusion, as well as the finding that Evaluative PCK negatively predicts math-related language outside of circle time, are discussed. (author abstract)