The aim of this thesis is a historico-comparative study of Zambian Plateau Tonga (Guthrie's M 64) and seven related lects (Valley, Toka, Ila, Lenje, Soli, Subiya and Totela). Some previous studies have treated Subiya and Totela as a distinct subfamily of Bantu (Guthrie's K 40), while others agree in attaching it to M 60; SI has also been associated with Lunda (Guthrie's L 51) and Luvale (Guthrie's K 14). The present study is based on wordlists of some 650 items Including Swadesh's 200-wordlist of basic vocabulary collected for each of the lects during a five-month field trip to Zambia in 1987. The study examines this data both synchronically (Chapter 2) and diachronically, attempting to reconstruct an inventory of Proto-Tonga consonant and vowel phonemes (Chapters 3) and relating this to Guthrie's Proto-Bantu (Chapter 5). Hierachically two broad subdivisions of the Tonga lects can be made Subiya and Totela together form one branch of Tonga as evidenced by certain shared innovations. The other branch groups together Plateau, Valley, Toka, Ila, Lenje and Soli on the basis of another set of phonological developments and the six lects are collectively referred to as Core Tonga. However, this division cannot be rigidly adhered to because Subiya and Totela to some extent participate in innovations affecting one or more members of the core group. It cannot be decided at the moment whether Sb and Tt together with the core lects form one distinct subfamily of Bantu or whether Sb and Tt form a different subfamily with some other lects not studied here. (Chapter 4). Some of the innovations link neighbouring lects and point to diffusion of phonological features across a geographical continuum. This convergence is further illuminated by the discussion of sociolinguistic factors in Chapter 6.