Over the past 30 years, numerous policies have focused on closing the achievement gap in the United States. Despite this emphasis on mastery and proficiency, students continue to perform below expectations. We use standardized math test performance to explore whether schools have been able to eliminate the achievement gap across middle-level education (students aged 12 to 14 in school years 6 through 8), and whether it's possible to maintain consistent improvement within a school. Additionally, we examine how schools work to improve performance of their lowest-performing students. We find that most schools have very little impact in increasing (or decreasing) the proportion of proficient students across the middle grades. Furthermore, schools that do produce consistent gains in proficiency are generally not the same schools that improve the performance of lowest-performing students. These patterns also hold across key student subgroups: African American, Hispanic, and low-income students. Results from this study bring into question the logic behind focusing on proficiency for all students as a measure of school success.