BACKGROUND: Alcohol disproportionately affects socially disadvantaged groups including Aboriginal and Torres Strait Islander Australians. METHODs to assess alcohol intake for disadvantaged communities need to be able to capture variable or episodic drinking. The ability of a seven-day diary to capture typical consumption for a predominantly Aboriginal sample has not been assessed. OBJECTIVE: One aim of this paper was to examine agreement between a seven-day retrospective diary and 'usual' drinking assessed by a modified version of the Alcohol Use Disorders Identification Test question 3 (AUDIT-3m; two questions). Other aims were to describe drinking patterns as reported in the seven-day diary. METHOD: In 2012, consecutive adults attending an Aboriginal Community Controlled Health Service completed a cross-sectional health risk survey on a touch screen laptop (n = 188). Alcohol consumption questions included the retrospective diary and AUDIT-3m. Agreement was assessed using weighted kappa analysis. RESULTS: There was good agreement between the two measures of consumption; however, the AUDIT-3m questions identified more current drinkers. Respondents who were drinkers (54%) tended to consume large amounts per drinking occasion: almost half (46%) of diary completers reported nine or more standard drinks on at least one occasion in the last week. CONCLUSIONS: The seven-day diary did not adequately capture variability in alcohol consumption common among this sample. Although the AUDIT-3m appeared acceptable, alternative approaches to assess usual or risky alcohol consumption, such as asking about specific drinking occasions, or allowing participants to respond in non-standard drink sizes, also need to be considered for indigenous and other disadvantaged communities.