This article employs computer-assisted methods to analyse references to Aboriginal and Torres Strait Islander people(s) and issues in a newspaper corpus about diabetes. The objectives are to identify both the frequency and quality of social representation. The dataset consisted of 694 items from 12 Australian newspapers in a five-year period (2013-2017). The quantitative analysis focused on frequency (raw/normalised) and range (number/percentage of texts). The qualitative analysis focused on the identification of semantic prosody (co-occurrence with negative/positive words and phrases) and on selective social actor analysis. The qualitative analysis also compared choices made by the press to language practices recommended in relevant reporting guidelines. Key results include that references to Aboriginal and Torres Strait Islander people(s) or matters appear to be extremely rare. In addition, newspapers' language choices only partially align with guidelines. References that do occur can be classified into four categories: a) references to [groups of] people and other references to identity; b) names of services, institutions, professions, roles etc; c) nonhuman nouns related to health; d) non-human nouns related to culture. Qualitative analysis of the word COMMUNITY suggests that newspapers for the most part do recognise the existence of different communities at a national level. However, analysis of all references to [groups of] people shows that the vast majority occur in contexts to do with negativity, therefore having a negative semantic prosody. More specifically, there is a strong association with mentions of a higher risk, likelihood, or incidence of having or developing diabetes (or complications/effects). In sum, Aboriginal and Torres Strait Islander people(s) and issues lack in visibility in Australian diabetes coverage, and are associated with deficit framing, which can be disempowering. To change the discourse would require both an increased visibility as well as changing the deficit lens.