The linguistic map of Amazonia presents a startling jumble of languages and language families. While some families – most notably Carib, Arawak, Macro-Jê, and Tupí– are distributed widely throughout the region, their spread is interspersed with many dozens of tiny, localized families and language isolates, particularly in the Amazonian periphery. At the same time, distributions of lexical, grammatical, and phonological features suggest that this linguistic patchwork is overlaid in places by contact regions, where multilingualism has fostered lexical and/or structural resemblances among languages. This complex distribution of languages and linguistic features presents many challenges to our understanding of Amazonian prehistory. How did Amazonia's language families arrive at their present distribution? Why did some families spread over huge distances, while others came to occupy only tiny geographical pockets or are limited to a single language? What kinds of interactions among peoples led to the formation of contact zones, and how are these regions defined? Complicating these questions further is the fact that very little is known about many Amazonian languages, and relationships among them are in many cases a matter of conjecture. This article surveys our current understanding of language classification and language contact in Amazonia, and addresses various perspectives concerning the implications of these relationships for Amazonian prehistory.