In the late 1950s and early 1960s, before he began work on the languages of New Guinea, Stephen Wurm undertook considerable fieldwork on languages of northern New South Wales and southern Queensland. His fullest materials were on Duungidjawu, spoken just to the northwest of Brisbane, and were recorded between 1955 and 1964. Wurm was generous in making his materials available to selected researchers, and in 1997, an arrangement was made with Wurm for Suzanne Kite to write an MA thesis analysing these materials. These consisted of tapes and transcriptions, with WurmÕs translations of these in his own shorthand, which only he could read. When he was in Canberra, Wurm would spend one or two afternoons each week going over these materials with Kite, explaining the shorthand and reviving his knowledge of the language. He had never written a draft grammar of Duungidjawu, but effectively had one in his head. It was hard to remember things exactly after a period of almost forty years and Kite sometimes mediated between what was on the tapes and WurmÕs explications during their collaboration. Stephen Wurm passed away in late 2001, after the thesis had been approved but before this work could be published. This is a slightly revised version of Kite's thesis. It comprises an invaluable record of the language of the Duungidjawu people, and through this of their traditions, customs and laws. It is the only substantial record of a language which differs in various respects from prototypical non-prefixing Australian languages. It has five vowels and a fair number of monosyllabic words. Pronouns and nouns referring to humans or to dogs have distinct case forms. Following the grammar sketch are all the texts recorded by Wurm and a full vocabulary and thesaurus. All Wurm's information was provided by Willie McKenzie, believed to be about eighty years old in October 1955. He died in 1965.