We used time-varying effect modelling of two very large samples of fourth-grade students (N[subscript reading] = 148,240, N[subscript mathematics] = 152,220) to investigate associations between adoption and over-time implementation of a de facto cap on special education service receipt and over-time likelihoods of disability identification from 2003 to 2017 for Texas students including those from historically marginalized communities. Following the cap's adoption, Texas students who are Black or English Language Learners (ELLs) were gradually less likely to have been identified as having disabilities than students in adjoining statues who are Black or ELLs in analyses adjusting for individual academic achievement, family economic disadvantage, school fixed effects, and other explanatory factors. Findings provide additional evidence of the cap's specific associations with disability identification disparities for demographic populations especially likely to have experienced violations of the Individuals with Disabilities Education Act's Child Find and free and appropriate public education requirements.