This paper reports on a 3-year, NSF-funded research and development project called ACED: Adaptive Content with Evidence-based Diagnosis. The purpose of the project was to design, develop, and evaluate an assessment for learning (AfL) system for diverse students, using Algebra I content related to geometric sequences (i.e., successive numbers linked by a common ratio). A key feature of the project was its dual focus on both learning effectiveness and accessibility. The project consisted of two studies. Study 1 (N = 268) experimentally evaluated the impact on learning of two ACED features: (a) elaborated task feedback and (b) adaptive sequencing of tasks. Study 2 examined the accessibility of ACED for students with visual disabilities (N = 4; two blind and two low-vision)--in a condition involving both adaptive sequencing and elaborated feedback--using a variety of accessibility features. Available features included voicing of test content (via synthesized and prerecorded audio), talking tactile graphics, font enlargement, screen magnification,and so on. Together these studies lay groundwork for the design of AfL systems that can enhance the learning of individuals with and without disabilities.