A large body of research has found that, by being better able to serve as cultural translators and role models, demographically-similar teachers can increase students' achievement. These studies have tended to focus on the role of race and gender similarities between student and teacher. This study is the first to examine the role of native language similarity. Using a nationally representative dataset from the United States which allows each student to be matched with two of his subject teachers, this study exploits variation in contemporaneous test scores and whether the student shares the same native language as the teacher across two different academic subjects, within-student, to identify the effect of being assigned to a linguistically-similar teacher. The effect is examined separately for students who are native Spanish-speakers and students who are native English-speakers. It finds that, unconditional on teacher ethnicity, assignment to a native Spanish-speaking teacher reduces the achievement of native Spanish-speaking students, particularly in Science and English. However, once differences in teacher ethnicity are controlled for, a native Spanish-speaking student does no worse or better on his test score when assigned to a native Spanish-speaking teacher than when assigned to a non-native Spanish-speaking teacher. For native English-speaking students, assignment to a linguistically-similar teacher has no impact on achievement. This finding applies regardless of whether teacher ethnicity is controlled for.