This study examined the design, implementation, and instructional effectiveness of three of America's most widely disseminated comprehensive school reform programs (the Accelerated Schools Project, America's Choice, and Successful for All) over a four year period that encompassed the school years 2000-2001 through 2003-2004. The authors' goal in presenting this study of three comprehensive school reform programs is to provide readers with some key insights into the larger workings of design-based school improvement. During the course of the study, data were collected in 115 elementary schools in every region of the United States, with more than 5,300 teachers, 800 school leaders, and 7,500 students and their families participating. The authors' findings suggest that design-based instructional improvement can come in many and varied forms. Their findings also show the benefits of moving beyond highly generalized conceptual frameworks for describing program designs, patterns of instructional management, and specific instructional practices in schools. Third, the authors argue that this study shows the benefits of looking closely at instruction when studying the process of instructional improvement. Finally, the authors argue that the results of "A Study of Instructional Improvement" (SII) confirm the larger "logic model" of school improvement by design that they described earlier in this report. (Contains 1 table, 1 figure, and 3 footnotes.) [This report is a reprint of a chapter that originally appeared as Chapter 49 of the "Handbook of Education Policy Research," edited by Gary Sykes, Barbara Schneider, and David N. Plank and published for the American Educational Research Association by Routledge Publishers in 2009. The reprinted chapter presents key findings from "A Study of Instructional Improvement," a study that was conducted under the auspices of the Consortium for Policy Research in Education and directed by Brian Rowan, David K. Cohen, and Deborah Loewenberg Ball (all at the University of Michigan).]