English learners (ELs) represent a large and rapidly growing proportion of US public school students. Federal law requires public schools to provide ELs with linguistic supports necessary to equalize their access to instruction with that of native English-speaking students. Most ELs are eventually reclassified and receive instruction within a general education classroom without supplemental language supports (Umansky and Reardon 2014; Kieffer and Parker 2016). Indeed, the snapshot statistic that about 10.3% of public school students nationwide are ELs (National Center for Education Statistics 2022) considerably understates the proportion of students exposed to EL services because it excludes those who entered the system as an EL and have since acquired sufficient facility with English that they were reclassified as Fluent English proficient. The process of reclassifying an EL has stakes for both the student and the school system. Removing language support services too soon would harm a student's growth, while delaying their reclassification can lead to stigmatization and limit the student's academic opportunities (Umansky and Dumont 2019). The shock from transitioning into a general education classroom itself may also pose challenges beyond those associated with the student's facility with English. From the state's perspective, providing supplemental language supports to students who no longer require them would unnecessarily strain resources and inflate demand for teachers licensed to instruct ELs, a persistent shortage area within the education workforce (Cowan et al. 2016). We apply an event study design to longitudinal administrative data from Indiana to shed new light on the effect of reclassification on later student outcomes. Across grades and subjects, we find a consistent pattern whereby student test scores trend upwards during years approaching reclassification, plateau or reverse during their first year as a non-EL, and then again trend upwards in subsequent years. The magnitude of the negative effect in the first reclassified year on student test scores is substantial, ranging from -0.103[sigma] to -0.300[sigma] across grades and subjects. The gains these students make when they return to a positive trend are not large enough to recover the losses from the initial reclassification year. Following the strategy recently developed by Rambachan and Roth (2023), we demonstrate that causal interpretation of our results is robust to allowing for very large violations to the parallel trends assumption. Though the pattern of our results clearly indicate a need for policymakers to identify strategies for improving student outcomes during their first reclassified year, they do not clearly point to the most effective strategy for achieving that goal. We can think of two potential strategies available to policymakers that rely on different assumptions for what factors drive the disruption in student outcomes during their first year as a non-EL. One potential interpretation of our results is that students are reclassified before they have achieved sufficient proficiency in English to succeed in a general education classroom, which would suggest benefits from increasing the standard for reclassification. However, we find it equally plausible that the immediate negative effect is driven by factors associated with the student adjusting to a new learning environment, which would suggest policymakers should focus on adopting strategies to ease student transitions into general education. Our results contrast with the findings from a substantial body of research. The several prior studies that apply a regression discontinuity (RD) design to measure the effect of reclassification typically find no significant impact on later student test scores (Robinson 2011; Robinson-Cimpian and Thompson 2016; Cimpian, Thompson, and Makowski 2017; Reyes and Hwang 2021; Onda and Seyler 2020; Pope 2016; Johnson 2020; Chin 2021), or the likelihood a student graduates from high school (Johnson 2019; Carlson and Knowles 2016; Cimpian, Thompson, and Makowski 2017). From a policy perspective, the null effects reported in prior RD studies have a promising interpretation: If school systems appropriately reclassify students when they no longer require linguistic supports we would expect them to perform similarly with or without EL services (Robinson 2011). However, a key limitation with applying this interpretation to the reclassification literature is that prior RD estimates are arguably under-powered. As in prior studies, we produce within our data RD estimates that lack sufficient power to detect meaningful effects. In most cases our null RD estimates do not significantly differ from our large but precisely-estimated event study results. We argue that this pattern of results suggests that the imprecise estimates from prior RD studies may also mask a true underlying treatment effect, raising questions about this seemingly robust finding within the literature. We encourage future work that applies alternative identification strategies to measure the effect of reclassification on student outcomes in other localities and contexts. Indeed, our analyses demonstrate the potential for researchers to use event study strategies to measure the effect of supplemental services on student outcomes more generally. The central concern with applying event study strategies in such a context is the likelihood that trends in student outcomes are associated with (and in some cases may directly trigger) the student's assignment to receive such supports. However, that it is illegal to randomly assign students access to such services and credible quasi-experimental designs are rarely available have left policymakers with little credible evidence for the impact of seemingly important services on the lives of a large and rapidly growing population of students. We present a case where the change in the trajectory of student outcomes associated with the timing of removing English language supports is substantial enough that selection bias alone cannot plausibly explain the results. We further demonstrate how the "HonestDiD" strategy allows the researcher to specifically quantify the magnitude at which the parallel trends assumption must be violated in order to explain the observed difference. The patterns we identify for student outcomes associated with reclassification as a non-EL are intriguingly consistent with Schwartz, Hopkins, and Stiefel (2021)'s recent event study analyses measuring the causal effect of assignment to special education services. We encourage future research employing alternative plausible identification strategies with transparent assumptions underlying causal interpretations to investigate the impact of EL and special education services on student outcomes.