A critical period for language learning is often defined as a sharp decline in learning outcomes with age. This study examines the relevance of the critical period for English-speaking proficiency among immigrants in the USA. It uses microdata from the 2000 US Census, a model of language acquisition and a flexible specification of an estimating equation based on 64 age-at-migration dichotomous variables. Self-reported English-speaking proficiency among immigrants declines more or less monotonically with age at migration, and this relationship is not characterised by any sharp decline or discontinuity that might be considered consistent with a "critical" period. The findings are robust across the various immigrant samples, and between the genders. (Contains 2 tables, 5 figures, and 10 notes.)