Texas has one of the highest populations of English Language Learners (ELLs) in the country, with a complex system for ELL identification, program placement, and high-stakes assessment. Spanish-speaking Latino ELLs represent a large proportion of this population in the state of Texas. The long-term academic achievement of ELLs identified in elementary grades and educated in different program placements is not well known. Prior research presents support for Bilingual Education models as most promising for future student achievement. Using strict sampling criteria and analysis of variance (ANOVA) procedures, this study compared student achievement during secondary years, as measured by high-stakes assessments, of demographically similar Latino former-ELLs, long-term ELLs, and non-ELLs who received a variety of language service programming during elementary school. Former-ELLs are students reclassified as English proficiency by the start of secondary education, while long- term ELLs are not reclassified until after 5th grade or not at all. Data was examined from a cohort of 18,188 students enrolled in all 10 major urban school districts in Texas from 3rd to 9th grades, between 2003-2004 and 2009-2010 school years. Math and reading high-stakes test (TAKS) scores during students' 7th and 9th grade years were used as outcome variables, while controlling for students' baseline English proficiency level.