This project examines upper elementary and middle school students' knowledge of scientific evidence. Informed by literature in cognitive development, learning sciences, and science education, this proposal argues that science educators have typically treated evidence as a simple and unitary construct that is disconnected from other phases of scientific activity. Evidence in the philosophy and history of science, on the other hand, is multifaceted, sophisticated, and involves the coordination of disciplinary knowledge and methodological practices. Based on a conceptual analysis of evidence in this literature, I developed a framework of evidential dimensions that important to scientific reasoning. Two fifth and two seventh-grade classrooms in suburban Midwestern public schools completed one of two science narratives containing a subset of dimensions from the framework. High and low performing students on the narratives were interviewed. Semi-structured interviews were conducted with participating teachers as well as classroom observations. Teachers also provided descriptions of inquiry tasks used in the classroom. Results indicate students struggled reasoning with and about aspects of evidence from the framework. Further, teacher interviews, classroom observations and tasks reveal oversimplified notions of evidence at play in the classroom, and I suggest these instructional facets are associated with student performance.