The development of science simulations and games has outpaced their grounding in theory and research on learning and assessment. Both genres draw upon the affordances of technology to provide students with opportunities to engage in active exploration and experimentation in virtual science environments that are impractical or impossible to access otherwise. Science simulations typically fall into three categories—simulations of science phenomena, multiuser virtual environments, and virtual laboratories. Educational, or “serious,” games immerse learners in realistic science worlds as the students gain and use knowledge to solve mysteries or to advance through multiple levels. Learning objectives may be acquisition of declarative knowledge—scientific facts, concepts and principles; formation of schema—connected knowledge structures about science systems; acquisition of procedural knowledge—steps in using tools and equipment; or strategic knowledge—learning when to employ inquiry skills and hone model-based reasoning. Even broader goals may target metacognitive strategies and epistemic strategies for far transfer. Simulations and games typically present tasks that are generally interactive, requiring the student to construct understandings and conduct iterative investigations within the virtual environments. Activities may vary from set procedures to graduated levels of complex strategies. Tasks may provide feedback and hints to scaffold learning progress. Relatively scarce, however, are clearly articulated descriptions of the evidence gathered to support claims of student learning. In most instances, rich streams of data from interactive tasks are not tapped as evidence of learning. Assessments of learning from simulations and games often resort to paper-based conventional task and item formats with limited possibilities for measuring the significant kinds of complex science learning targeted.