A comprehensive review of the literature was conducted for articles published between 1985 and May 2009 to (a) examine the degree to which science content was taught to students with severe developmental disabilities and (b) and evaluate instructional procedures in science as evidence-based practices. The review was organized by a conceptual model developed for science content. Seventeen experiments were analyzed for research quality where science content was taught to this population; 14 of these studies were viewed to be of high or adequate quality. In general, we found systematic instruction as an overarching instructional package to be an evidence-based practice for teaching science content. Furthermore, components of systematic instruction (i.e., task analytic instruction and time delay) were analyzed. We discuss the outcomes to reflect how to teach science, what science content to teach, why to teach science, and recommendations for future research and practice. (Contains 2 tables and 1 figure.)