Background/Context Sustained school improvement requires adequate organizational and instructional coherence, yet, in typical high schools, subject department organization, norms of teacher professional autonomy, and involvement in multiple initiatives present powerful obstacles to forging a coherent approach to improvement. This study examines the relationships among coordination, coherence, and improved achievement in five New Zealand high schools voluntarily participating in an initiative to improve the achievement outcomes of underserved student groups. Purpose The first purpose of this study is to develop a theoretical framework for describing and evaluating coordination and coherence and to apply it to the implementation of an improvement initiative. The second purpose is to examine the relationship between coherence and longitudinal trends in pass rates on national qualifications. Research Design Adjusted odds ratios were used to compare the pass rates of students in the five high schools before and after the intervention. This analysis, conducted independently of the study of school coherence, yielded a group of schools whose pass rates had improved and a second group that had maintained stable pass rates through the course of the intervention. Using case studies written for each of the schools, we systematically compared the schools in each group on degree of coordination in five domains of activity and in overall coherence. Findings There were considerable differences in degree of coordination between the two groups in three of the five domains. Improving schools were more tightly coordinated in how they organized to achieve results, in teacher professional learning, and in strategic leadership. Principal capability and willingness to address barriers to improved achievement, which was part of strategic leadership, was markedly greater in the improving than in the maintaining schools. In the remaining two domains, goal-setting and teacher culture, differences were smaller but still in the expected direction. The overall coherence of each improving school was independently evaluated as high, while the maintaining schools were evaluated as low or medium in overall coherence. Conclusions We discuss the implications of coordination and coherence for the successful implementation of any school improvement initiative. We argue that, since educational work contains weak cues about the adequacy of the performance of any component activity, we cannot rely on adequate coordination emerging through the interactions of interdependent agents. More emphasis must be given, in coordination theory and improvement practice, to steering such interactions through task design, widely distributed expertise, and strategic leadership.