A main goal of science education is to help students learn to reason scientifically. A main way to facilitate learning is to engage students in inquiry activities such as conducting experiments. This article presents a theoretical framework for evaluating inquiry tasks in terms of how similar they are to authentic science. The framework helps identify the respects in which these reasoning tasks are similar to and different from real scientific research. The framework is based on a recent theory of reasoning, models-of-data theory. We argue that inquiry tasks commonly used in schools evoke reasoning processes that are qualitatively different from the processes employed in real scientific inquiry. Moreover, school reasoning tasks appear to be based on an epistemology that differs from the epistemology of authentic science. Inquiry tasks developed by researchers have increasingly captured features of authentic science, but further improvement is still possible. We conclude with a discussion of the implications of our analysis for research, assessment, and instruction.