During a two-year span, a cohort of K-8 science teachers undertook the task of creating formative assessments aligned with the Next-Generation Science Standards for their individual classrooms. Practice-based professional development was utilized to support teachers in a cyclical manner as they cooperatively developed, administered, and reflected on a series of assessments they designed across two study phases. In order to measure developmental progression of the formative assessments across the study's two phases, project leadership designed an assessment quality rating instrument targeting four components of a given formative assessment: (a) learning goal(s), (b) task/prompt, (c) success criteria, and (d) next instructional steps. Using this criterion-based instrument, 123 assessments were subjected to consensus scoring using a 0-3 rating for each component (e.g. learning goals). Results indicated teachers developed higher quality formative assessments during the project's second phase. More specifically, they developed learning goals and success criteria that were identified as being multi-dimensional, meaning the assessment component was designed to assess more than content. Implications for practitioners and researchers are discussed, specifically related to the supports needed for teachers learning to progressively design and implement full spectrum, science-based, formative assessments. Finally, suggestions for future research are discussed given the potential value of using assessment design to introduce a cycle of teacher-driven self-reflection.