Purpose: Supporting teachers' use of data has become a large part of educational leaders' instructional leadership. Drawing on sensemaking theory, we explore how features of data and teachers' perceptions of them may matter when teachers consider student learning data. Design: The article draws on a 1-year, comparative case study of five low-income, high needs middle schools in three districts. Data sources include interviews with district leaders (n = 13); school administrators, coaches, and case study teachers (n = 73); focus groups (n = 6) with non--case study teachers (n = 24); observations (n = 20); web-based activity logs; and document review. Findings: Different forms of data lead to a wide range of instructional responses. State assessments, used largely at the beginning of the school year, aided teachers in grouping their students but did little to promote change in instructional delivery. District benchmark data, which teachers did not generally find useful, were associated with reteaching and retesting content, creating small groups, and having students reflect on their data, usually without a shift in pedagogy. Data from common grade assessments--valued for their closeness to instruction as well as their predictive information for future success on state assessments--were most often tied to regrouping or reteaching standards not yet mastered by students, but there was some promise of change in pedagogy. Classroom assessment results and student work, identified by teachers as very useful, were proportionately most often linked to changes in instructional delivery. Implications: We offer implications for leadership preparation and practice, education policy, and future research on data use.