Observation measures of instructional quality tend to fall into two broad categories--those for use across subject areas and those intended for use in specific subject areas. The move toward content-specific measures is a result of research suggesting that effective teaching looks different across subject areas and that both content knowledge and pedagogical content knowledge are critical for effective teaching. Choosing either across-subject and subject-specific observational evaluation measures has interesting implications for beginning teachers who often struggle with global instructional practices like establishing a positive classroom climate and classroom management, while concurrently implementing rigorous content-area instruction. Despite these challenges, evidence has found that beginning teachers' effectiveness at increasing student achievement improves during the first few years on the job. Yet, no research has examined whether beginning teachers are more likely to show improvement on a content-focused measure or a more general measure or whether improvements on one type of measure are associated with improvements on the other. These questions are important given the emphasis placed on evaluation of teachers' instructional practices as means for improving overall teacher quality. This study will answer the following research questions: (a) To what extent do beginning teachers improve their instructional quality during their first three years of teaching, as measured by a standardized observation measure of global classroom quality, the Classroom Assessment Scoring System (CLASS), and a subject-specific rubric in mathematics, the Instructional Quality Assessment (IQA)?, and (b) Do teacher instructional quality ratings from the IQA and the CLASS, similarly identify high quality and low quality middle school mathematics teachers during their first three years of teaching? While the teachers' IQA ratings did not change over time, teachers improved on the emotional support and classroom organization dimensions of the CLASS. Findings also show that teachers who exhibit higher levels of emotional support and classroom organization during also rate at higher levels of the math-specific IQA, indicating that teachers with stronger relationship with students and classroom management may be more be likely to employ reform math ideals (Contains 3 figures and 5 tables.)