Background/Context: The emphasis on scientific practices articulated by the National Research Council framework and the Next Generation Science Standards requires significant pedagogical shifts for U.S. science teachers. Purpose/Objective/Research Question/Focus of Study: This study provides a rare window into the challenges and opportunities teachers encounter as they introduce argument writing into their science classrooms with support from the National Writing Project's Inquiry into Science Writing project. The purpose of this study is to better understand the teacher-change process so as to inform the development of future professional development efforts. Population/Participants/Subjects: Case studies were drawn from a professional development network led by the National Writing Project to support teachers in studying and improving their practice while sharing knowledge and benefiting from the expertise of others. The network included 28 middle school teachers at five writing project sites around the United States; the case studies presented in this article are based on the experiences of three of these teachers. Intervention/Program/Practice: The Inquiry into Science Writing Project was a 2-year practitioner-driven professional learning experience seeking to better understand and support student practice around evidence-based science writing. During the duration of the project, teachers taught at least one lesson series culminating in written arguments by students each semester, and participated in two summer institutes, an ongoing national professional learning community, and monthly meetings of their local teacher research group. Research Design: The study uses a qualitative comparative case study approach. Data Collection and Analysis: The case studies draw on interviews, lesson artifacts, written teacher reflections, and samples of student work. Conclusions/Recommendations: The study findings reinforce the complexity of the change process: The relationship between teachers' knowledge, beliefs, and attitudes and their practice was not linear and unidirectional (i.e., change in attitude leads to change in practice) but rather iterative and mediated by both student work and the external supports they received. These findings confirm the need for sustained learning environments with features that promote enactment and reflection on student work to support teacher change. Further, they suggest that professional development providers should think about how to build habits of reflection into their own design processes, allowing space for feedback and learning from practitioners.