Background/Context: Educational researchers and policymakers have often lamented the failure of teachers to implement what they consider to be technically sound assessment procedures. In recent years, the belief that teachers are unwilling or unable to implement appropriate assessment procedures has contributed to the rapid expansion of high stakes, standardized testing in schools. Supporters of No Child Left Behind (NCLB) have contrasted teachers' assessment practices with standardized testing, arguing that teacher-created classroom assessments lack the technical characteristics required to produce trustworthy measures of student learning or compare large populations of students. Research Question/Focus of Study: Through a case study of New York City's Central Park East Secondary School (CPESS), in the years when it served as a model for progressive American school reform, Duckor and Perlstein demonstrate the usefulness of an alternative to reliance on the technical characteristics of standardized tests for constructing and judging assessments: teachers' self-conscious and reasoned articulation of their approaches to learning and assessment. Research Design: In order to determine CPESS teachers' assessment practices and the process through which they were developed, Duckor and Perlstein conducted semi-structured oral history interviews with a sample of CPESS teachers. They triangulated teachers' recollections through a content analysis of course assignments, rubrics, grading reports, and other artifacts of assessment at CPESS. The sources of this data included published accounts of CPESS and primary sources provided by teachers or uncovered in archival research. Conclusions/Recommendations: Duckor and Perlstein conclude that when teachers are given opportunities for genuine, shared reflection on teaching and learning and classroom practices are tied to this understanding, fidelity to what they call the logic of assessment offers a more promising framework for the improvement of schooling than current forms of high-stakes, standardized accountability. Thus, instead of expecting teachers to rely on data from standardized assessments or replicate features of standardized testing in their own assessment practices, researchers, policymakers and teacher educators should promote fidelity to the broader logic of assessment.