The purpose of this article is to describe the vexing issues that arise as researchers try to define and assess the knowledge teachers need to teach reading and writing effectively in the primary grades. Over the past several years, the authors of this article have developed and tested an assessment system of teacher knowledge under a grant titled "The Primary Grade Reading & Writing Teacher Knowledge Project." The research team developed and tested 2 subscales: a traditional multiple-choice test that measures the inert or "in the head" knowledge teachers possess about reading and writing instruction, and an accompanying classroom observation instrument that measures the "enacted" or "classroom teaching" "knowledge" teachers use to teach reading and writing to young students. This article reports on 6 conceptually and methodologically vexing issues that arose as the authors developed and tested the assessment system. Issues revolved around the specific knowledge to measure, the evidence needed by various stakeholders as convincing evidence of that knowledge, potential concerns related to the "use" of measures of teachers' knowledge, queries about the best way to measure that knowledge, special problems the use of classroom observations present for measuring teacher knowledge, and the predictive validity of such knowledge measures. (Contains 1 figure.)