Purpose: Currently under the No Child Left Behind Act, cut scores of successive cohorts of students are used to identify schools that are successful and others that are failing. For many schools, however, this score does not provide an accurate, fair, or comprehensive performance assessment. The focus of this study is to compare three alternative approaches for estimating student progress with respect to their accuracy, equity, and usefulness in making inferences about school effectiveness. Research Design: The study compares achievement estimates compiled across a 4-year period from successive and longitudinal cohorts in a multilevel sample of 123 elementary schools and several thousand students. Findings: First, method effects associated with the designs interact with estimates of school effectiveness. Less than 20% of the schools are similarly identified as meeting performance standards across all three designs investigated. Second, in each approach, contextual indicators (e.g., student composition) dominate in explaining between-school differences in proficiency levels. Third, the longitudinal-cohort approach provides more comprehensive results about performance because of the simultaneous focus on student proficiency and growth. Equity and usefulness are also increased in comparisons of student growth because student composition and context effects are diminished and, conversely, process effects (e.g., academic expectations) are increased in explaining school differences. Conclusions: The comparisons suggest caution in making decisions about schools based on proficiency levels because both method effects and extraneous variables interact with the validity of estimates obtained. Modeling students' growth trajectories, however, can uncover positive school results that are missed in the current proficiency-status approach. (Contains 2 figures, 5 tables and 14 notes.)