The number of students that live in families where a language other than English is spoken has risen relative to the English-only households in the United States over the last 25 years. These students face the dual challenge of mastering English while acquiring academic skills and knowledge. The education of these students has been shaped by several national- and state- level legal decisions of the past that have required schools to take action by providing services to help linguistic minority students overcome language barriers that impede their equal participation. Those services most often take the form of either an "English-immersion" approach, in which students receive all instruction in English, or employ a "bilingual model" where students are initially taught in some combination of English and their native language and eventually transfer to English-only classrooms. The question of which model of instruction is more "effective" has been notoriously difficult to answer and remains an open and controversial debate. The authors of this article take advantage of a policy passed in California that changed the default instructional program for ELs . Prior to 1998, most ELs were placed into bilingual education. This state-level legislation forced schools to move students to move to a different instructional setting--a change that would not had been chosen otherwise--providing a natural experiment opportunity to evaluate the effect of bilingual education versus English immersion on the academic achievement of ELs. Also analyzed is how changes in the composition of the classroom affect the academic achievement of students. Findings indicate that the achievement scores of ELs declined by the switch from bilingual education to English immersion programs, with the exception of grades 2-3 reading scores. The authors also found that there is a beneficial effect of placing ELs into classrooms with more native English speakers. Tables and figures are appended.