This article reports on a design-based implementation research (DBIR) project that addresses the question: How can classrooms be supported at scale to achieve the threedimensional learning goals of the Next Generation Science Standards? Inherent in this question are three key design challenges: (i) three-dimensional learning-the multidimensional changes in curriculum, assessment, and instruction required for three-dimensional learning; (ii) scale-the necessity of change at multiple scales in educational systems; and (iii) diversity-achieving rigor in our expectations with responsiveness to the enduring diversity of our students, classrooms, and schools. We discuss findings from the Carbon TIME project, which focuses on teaching carbon cycling and energy transformations at multiple scales. Findings focus on design and knowledge building in three interconnected contexts. (i) Assessment-understanding and assessing students' three-dimensional learning. Learning progression frameworks provide insight into students' reasoning and the basis for efficient and reliable classroom and large-scale assessments that have used automated scoring of constructed responses for over 80,000 tests. (ii) Classrooms-classroom discourse and learning communities. Six Carbon TIME units are based on an instructional model that scaffolds students' engagement with phenomena as questioners, investigators, and explainers. The units support substantial learning and reduce the achievement gap between high-pretest and lowpretest students, but with substantial differences among