Spanish language tests of 801 Cuban and Mexican immigrants showed no evidence of language loss during 50 years of U.S. residence; a few years after immigration, their English vocabulary approximated that of English monolinguals. The critical-age hypothesis was not supported for the acquisition of English vocabulary when English schooling and language usage were controlled by multiple regression. Most Ss continued to speak about as much Spanish as English; but read, wrote, and heard (on television and radio) far more English than Spanish. Under these conditions, Ss maintained Spanish dominance on tests of vocabulary recognition, lexical decision, and oral comprehension. Dominance was task specific and shifted to English on a category generation task about 12 years after immigration. No evidence of bilingual language interference was found; this is attributed to the strong Spanish foundation of the participants.