This mixed-methods study examined the extent to which teacher performance and student performance measures correlated, and to understand which specific practices of mathematics teachers in Grades 3-5 related to student performance. Research was conducted at five elementary schools in a large, urban north Texas school district. Data sources included component scores and recorded evidence from observation rubrics, interviews with campus administrators, and value-added modeling (VAM) student growth scores. Findings indicated a modest relationship between teacher performance levels and student performance levels. Lack of access to individual teacher VAM data, per district policy, might have impacted the strength of the relationship. Interviews with administrators and an examination of the evidence cited in the observation rubrics identified specific practices associated with highly rated mathematics teaching. Differences in administrators’ experience levels with both mathematics instruction and the observation instrument might have influenced rubric scores and the level of specificity shown in evidence statements.