Introductory biology courses are supposed to serve as gateways for many majors, but too often they serve instead as gatekeepers. Reliance on lectures, large classes, and multiple-choice tests results in high drop and failure rates. Critiques of undergraduate science education are clear about the problems with conventional introductory science courses, and yet the problems persist. As David Hanauer and Cynthia Bauerle explain, "Given the potential for science to address important problems, undergraduate programs ought to be functioning as busy portals for engaging students' innate fascination and developing their understanding of the nature and practice of science. Instead, recent studies suggest, the opposite is true: over half of the students who enter college with an interest in science do not persist in their training beyond the first year or two of introductory coursework." Researchers and expert practitioners have long proposed using student-centered, active learning strategies to improve engagement, learning, and achievement. Others have documented the ways class size is important for student-centered pedagogy. Following Hanauer and Bauerle, who recommend using assessment reform to facilitate such curricular innovations, the authors contend that the better the assessment and the more focused the guidance provided to the instructor, the greater the leverage. This article presents findings from a pilot study suggesting that authentic assessment embedded in best teaching practices can show what kind of change is needed. The study allowed the authors to observe the relative impacts of both class size and the use of writing as an assessment strategy, and thus to identify the sequence reform efforts must take. The purposes of this article are, first, to report on the experience responding to Hanauer and Bauerle's call, and second, to identify the key components that gave that "reform lever" additional power: the careful selection and preliminary testing of essay questions requiring critical thinking, the reduced size of one section of an entry-level biology course, the support of a networked improvement community, and guidance for the instructor during the testing of new methods