Federal legislation and policy increasingly seek to address minority overrepresentation in special education due to concerns that U.S. schools are misidentifying children as disabled based on their race or ethnicity. Yet whether and to what extent this is occurring is currently in dispute. We estimated racial disparities in disability identification using very large (e.g., Ns = 183,570, 165,540, and 48,560) student-level, nationally representative data sets and multivariate logistic regression including school fixed effects models along with tabulations of percentage with a disability among racial or ethnic groups across academic achievement deciles. Among children who were otherwise similar in their academic achievement, poverty exposure, gender, and English language learner status, racial or ethnic minority children were consistently "less" likely than White children to be identified as having disabilities. Minority children's disability underidentification was evident (a) in elementary, middle, and high school; (b) across racially diverse groups and specific disability conditions; and (c) throughout the achievement distribution. Contrary to federal regulatory and policy efforts, minority children have been less likely than otherwise similarly achieving White children to receive special education services in the United States since at least 2003.