Students with significant cognitive disabilities present an assessment dilemma that centers on access and validity in large-scale testing programs. Typically, access is improved by eliminating construct-irrelevant barriers, while validity is improved, in part, through test standardization. In this article, one state's alternate assessment data were analyzed to determine the impact of (a) administration supports based on students' level of independence and (b) a scaffold test administration format. Using structural equation modeling, we tested the extent to which students' level of independence mediated the relation between disabilities and latent content knowledge scores. We then tested the invariance of the measurement model across administration formats. The results provide evidence that these supports help students access the test without compromising the validity of test-based inferences.