Glossary and reading aloud test items are often listed as allowed in many states' accommodation policies for ELL students, when taking states' large-scale mathematics assessments. However, little empirical research has been conducted on the effects of these two accommodations on ELL students' test performance. Furthermore, no research is available to examine how students use the provided accommodations. The present study employed a randomized experimental design and a think-aloud procedure to delve into the effects of the two accommodations. A total of 605 ELL and non-ELL students from two states participated in the experimental component and a subset of 68 ELL students participated in the think-aloud component of the study. Results showed no significant effect of glossary, and mixed effects of read aloud on ELL students' performance. Read aloud was found to have a significant effect for the ELL sample in one state, but not the other. Significant interaction effects between students' prior content knowledge and accommodations were found, suggesting the given accommodation was effective for the students who had acquired content knowledge. During the think-aloud analysis, students did not actively utilize the provided glossary, indicating lack of familiarity with the accommodation. Implications for the effective use of accommodations and future research agendas are discussed. Three appendices are included: (1) Example of Read-Aloud Script; (2) Glossary Terms Used in Math Test; and (3) The Five Think-Aloud Items. (Contains 23 tables, 3 figures and 4 footnotes.)