This study developed a comprehensive measure of the stringency level of NCLB states' accountability systems, including the strength of their annual measurable objectives, confidence intervals, performance indexing, retesting, minimum subgroup size, and the difficulty levels of proficiency standards. This study related accountability stringency in 2003 to student achievement and achievement gaps on NAEP math and reading tests from 2003 and 2005. The results were inconsistent across grades, subjects, and ethnic groups. An increase in accountability stringency, such as requiring faster progress, not allowing for retesting, and allowing a smaller minimum subgroup size were related to improved math achievement for fourth-grade Hispanic students. Not using confidence intervals was related to higher math and reading achievement for White and Hispanic students. However, accountability stringency did not have any positive effects and even negative effects on eighth-grade Black students. (Contains 9 tables and 2 notes.)