Researchers are concerned that English language learners (ELL) may remain classified too long and, therefore, may not receive appropriate mainstream educational services. In this study, we investigate the effects of language classification on student outcomes in one California school district. Our ordinary least squares regression estimates demonstrate that students who become reclassified fluent English proficient (RFEP) in middle school have higher achievement scores and better behavioral outcomes. However, regression discontinuity models show that reclassification has no effects on students' outcomes. The findings of this study suggest that selection bias accounts for the positive associations between RFEP and desirable student outcomes. ELL and RFEP students have unmeasured characteristics that highly correlate with their language classification, therefore, their educational outcomes differ.