The purpose of this dissertation was to investigate the role of background, experience, and interactions with colleagues and students in the development of mathematical proficiency for teaching, operationalized as teachers content knowledge for teaching and beliefs about learning and mathematics. I conducted three studies in different contexts. The purpose of the first study of Texas K12 mathematics teachers in their first 5 years of teaching was to describe how mathematical proficiency for teaching multiplicative reasoning varied across preparation, school contexts, and a wide range of grade levels. Surprising findings from the first study led to two follow-up studies: (1) content knowledge for teaching was not positively related to length of teaching experience and (2) the length of student teaching had no significant relationship to mathematical proficiency for teaching. The second study used a longitudinal design to study change in Grades 68 teachers mathematical proficiency for teaching multiplicative reasoning topics. I found that teachers content knowledge for teaching increased over the semester, especially for teachers with less mathematical preparation, but that their self-efficacy beliefs decreased. In the third study, I found that the quality, timing, and length of student teaching were significant predictors of content knowledge and beliefs. For all three outcomes, practicum length was found to moderate the effect of timing, with early timing of student teaching having significant positive effects for prospective teachers in programs with shorter student teaching. These results will support future work designing and investigating interventions that support teachers on-the-job development of mathematical proficiency for teaching