Standards-based legislation passed in the last decade mandates the inclusion of English-language learners, or students with limited English proficiency, in large-scale, content-based assessments. Concerns emerge, however, regarding the validity of inferences for these populations based on standardized test results. This article examines selected contextual variables, their impact on test performance, and how the variables interact with the type of test accommodation (e.g., linguistically modified English and Spanish translation) that a student receives. Data from 849 8th-grade students in southern California suggest that numerous student- and classroom-level variables are associated with mathematics test performance, particularly the students' English reading proficiency. Students also perform best when the language of the mathematics assessment matches the students' language of mathematics instruction, holding constant key student and classroom-level variables. Methodological issues related to accommodations research with English-language learners are discussed.