The No Child Left Behind Act requires schools to begin testing new English Learners (EL) in English language arts within three years after they enter school and holds schools accountable for their performance on these exams. Yet very little empirical work has examined exactly how long it takes EL students to become proficient in English and how the time to proficiency varies for different types of students. Linguistic theorists suggest, for instance, that the age at which students begin learning a second language may substantially influence their probability of obtaining proficiency quickly. Using panel data on English Learners (EL) in New York City public schools, I examine how long it takes students to become minimally-proficient in English and how the time to and probability of proficiency differs for students by their age of school entry. Specifically, I follow four entry cohorts of ELs and use discrete-time survival analysis to model variations in the rate at which different age groups acquire proficiency. I find that approximately half of the students become proficient within three years after school entry but that age of entry lowers the speed with which children can become proficient. Age of entry differences are robust to controls for differences in other student characteristic and the schools they attend. The results suggest that federal, state, and local policies regarding the testing of EL students in academic English should consider more flexible time limits. (Contains 6 tables.)