Christoph and Nystrand (2001) argue that teachers need to take pedagogical risks in order to realize a transformation from teaching monologically, where it is ultimately only the teacher's voice that matters, to teaching dialogically, where multiple voices collide to foster learning. The pedagogical risks of teaching dialogically include sacrificing a large measure of teacher control of the conversation and not knowing what will be said or even exactly what will be learned in advance. Particularly in a standards-driven environment--where prespecified outcomes are what are presumed to count and matter--these risks are substantial. The authors begin this article about "teacher learning" of dialogically organized reading instruction with the assertion that these risks are worth taking. There is substantial evidence that children and adolescents learn more in dialogic environments and have a deeper understanding of what they read, evidenced on end-of-year tests and even on standardized measures of achievement such as the California Achievement Test 6. But looking beyond these measures for student performance, the authors believe there are strong philosophical reasons for establishing classroom contexts in which students' voices shape the direction of subsequent learning. There is evidence that students learn different epistemological assumptions about literacy and knowledge in classrooms that are dialogically versus monologically organized. Within the specific context of a course devoted to the learning of dialogically organized instruction called The Discourses of Teaching Reading, teacher participants were asked to explore dialogically organized instruction through analysis of their own and their students' language during reading discussions facilitated twice a week with elementary-aged students. In this article, the authors present the findings and insights from this study. (Contains 1 figure.)