The general aim of the present project is a historical-comparative study of two of the largest language families of the South American lowlands: Cariban and Tupian ('Katu'). Although the genetic relationship of these two families has not yet been universally accepted, Rodrigues (1985) presented some very promising lexical and grammatical evidence for grouping them together. In Rodrigues (2000) a further genetic relationship of Cariban and Tupian with the Macro-Jé language stock is suggested. These ambitious proposals, involving a very substantial part of the native languages of South America, have to be investigated systematically. In the present project we intend to give attention to the first proposal, which will be approached in the following way: (1) establishment of a computer data base of the available data on the Tupian and Cariban languages; (2) evaluation of gaps in the data and selection of target languages for further data collection; (3) field research involving these target languages; (4) application of the historical-comparative method to the data collected within each family; (5) comparison of the reconstructed elements, so as to establish a basis for a possible genetic relationship of the two language families. The final result will be a monograph dealing with the reconstruction of the Cariban-Tupian stock, which will expectedly increase our knowledge of the history of human occupation of the South American lowlands. The project will be part of a comprehensive programme dealing with the description, the reconstruction and the genetical connections of South and Middle American languages in general.